= Cambridge University Undergradaute Research Opportunities Programme - UROP

UROP Projects

The UROP is closed for 2021. New projects will be added in 2022.


Application Restrictions EPSRC funded Projects
The UROP is designed to support undergraduates studying at the University of Cambridge who are going to return for at least one more year of undergraduate study.
Final Year undergraduates and Postgraduate students should not apply.

For projects that are EPSRC funded and marked with the blue flag,  European Union citizens ONLY students must meet certain criteria to apply.

( See LINK for criteria )

For EPSRC funded projects, students should be in the middle years of a first degree within EPSRC's technological remit.


Information for Cambridge University students


Information for Cambridge University Staff

Please report any broken links to UROP Web Page Co-ordinator

Ms Vicky Houghton


Magic Air

Lead Supervisor: Dr. Adam Boies of the Department of Engineering
Project Taken

Project Description:

As part of the EPSCR Grand Challenge project MAGIC, a two-week intensive field study was carried out in London in September 2019 collecting a wide range of data including air pollution measurements, meteorological data and video footage of the traffic. The study was done in conjunction with TfL who changed the signal timings at a junction during the study period to understand the impact of different signal timings on vehicle emissions, overall air pollution and pedestrian exposure.

Understanding the composition of the traffic plays a key part in understanding the influence of traffic on air quality. The student's project would include the development of an algorithm to extract number plate information from the video footage, match this number plate information with the available vehicle information and apply an instantaneous vehicle emissions model.

  • The project will last for 10 weeks.
  • The ideal candidate will have good knowledge of Python and has experience using Computer Vision. Good communication and team working skills are also required.
  • The project can be carried out remotely.

  • Please submit your cover email with CV attached to Dr. Adam Boies.

    Insertion Date: 13 July 2021


    (Back to List)

    Modelling global land cover change and implications for carbon balance

    Supervisor: Dr. Andrew Friend, Department of Geography
    Project Taken

    Project Description:

    This work will utilise a global vegetation/soils model being developed for the TRENDY project (https://sites.exeter.ac.uk/trendy/). The implications of land use change 1901-2020 on the global carbon balance are an important component of this work, but still require consistent implementation within the overall model framework. The student can either (i) develop new model approaches to implement land use management within the global model framework (the management data are already available, but their input not yet incorporated – this requires careful consideration of land cover dynamics, changes between states, and conservation of quantities), or (ii) develop new model approaches to how management affects particular land cover types, such as rice paddies, maize fields, forest plantations, etc. (this will require consideration of timings of planting/harvest, crop rotation, decomposition, effects of extreme weather events, etc.). In terms of work outwith modelling, the student could visit local agricultural institutes, such as NIAB, to exchange ideas and gather data, and/or visit other labs in the TRENDY project to discuss implementations.

  • A background in quantitative methods/maths/physics/modelling or environmental science/biology is essential. Ideally experience of a programming language such as Fortran and/or Python.
  • Full project details and eligibilty criteria are available on the Cambridge NERC Doctoral Training Partnerships website.
  • The duration of REPs will be eight weeks over the summer holiday period (mid-July to September).
  • If the project has the ability to be carried out remotely, it will be noted in the project details link above.
  • Deadline for applications, including one academic reference, must be received by Wednesday 30 June 2021.

  • Insertion Date: 11 June 2021


    (Back to List)

    Flax biocomposites for car headliners

    Lead Supervisor: Dr. Darshil Shah of the Department of Architecture
    Industrial Collaborator: Dr Laurent Mougnard (Howa-Tramico SAS, France)
    Project Taken

    Project Description:

    The ERDF-funded FLOWER project aims to develop low-cost, locally-produced flax fibre reinforcements for the composite industry. The outcomes of this project will enable to achieve the aspirations of using sustainable, cost effective and environmentally friendly lightweight composites as a viable alternative for automotive, advertising and sailing sectors.

    In this 3-week research internship, you will test biocomposite materials, principally flax-fibre reinforced sandwich structured composites, as potential replacements to currently-used glass fibre reinforced sandwich structure composites, for interior headliners in cars. Specifically, you will test the thermo-mechanical properties and damping behaviour of the materials, for instance using a Dynamic Mechanical Analyzer (DMA). There may also be opportunities to complete some X-ray microtomography examination. You will be asked to meticulously write down the methods you use, and analyse the results you obtain into a format of a report.

  • Suitable for a student interested in materials (natural and polymer composite materials) and structural engineering.
  • General workshop skills, experience of mechanical testing, and some analytical modelling skills are recommended.
  • The proposed length of this project is for a continuous 3 week period, sometime between mid-August and early-October.
  • The funding is part of the ERDF-funded FLOWER project, awarded to the Centre for Natural Material Innovation. No restrictions to candidates.
  • We especially welcome applications from historically marginalised people.

  • Please submit your cover email with CV attached to Dr. Darshil Shah.

    Insertion Date: 13 July 2021


    (Back to List)

    How does chemical diversity beget biological diversity?

    Supervisor: Dr. Andrew Tanentzap, Department of Plant Sciences
    Project Taken

    Project Description:

    Resource competition shapes the functioning of ecosystems by influencing biodiversity. Available resources determine if species compete for common resources, facilitate each other by modifying resources to increase their availability, or avoid interacting because of resource (niche) separation. In bacterial communities that control important ecological functions like carbon cycling and water quality, resource competition is structured by the capacity of species to use dissolved organic matter (DOM). However, a key unknown is how these interactions change with the availability of resources. Recent advances in mass spectrometry have now enabled the identification of thousands of molecular formulas in DOM, thereby providing a clearer understanding of resource dynamics than ever before.

    The objective of this studentship is to understand how biotic interactions change with resource availability, specifically the chemical diversity of DOM. Our hypothesis is that different resource composition results in different interaction patterns. We expect more antagonism when bacteria grow on labile resources and more mutualism/exploitation when they are grown on more recalcitrant resources. To test our hypothesis, you will grow ten well characterized environmental bacteria on different DOM sources. Comparing how species grow in pairs vs monoculture is a standard approach to estimate interactions. However, these estimates often come from growing bacteria on artificial media with an overly simplified resource composition that do not reflect nature.

    In weeks 1 to 2 of this project, you will collect water from around 10 sites in Cambridge and use solid phase extraction to isolate DOM. The composition of this DOM will be characterised with UV spectrophotometry and mass spectrometry. In weeks 2 to 5, you will create growth media with this DOM and grow pairs of bacteria in 96 well plates. You will determine total productivity of colonies by with spectrophotometry and cell counts. Comparing growth alone with growth in pairs, corrected by frequencies, will allow you to calculate interaction coefficients. In weeks 6 to 7, you will review literature on interaction coefficients and test different methods to analyse your data. Finally, in week 8, you will test how interaction coefficients vary among taxa and with resource composition.

  • Full project details and eligibilty criteria are available on the Cambridge NERC Doctoral Training Partnerships website.
  • The duration of REPs will be eight weeks over the summer holiday period (mid-July to September).
  • If the project has the ability to be carried out remotely, it will be noted in the project details link above.
  • Deadline for applications, including one academic reference, must be received by Wednesday 30 June 2021.

  • Insertion Date: 11 June 2021


    (Back to List)

    The contributions of climate change and evolutionary history towards dark diversity

    Supervisor: Dr. Andrew Tanentzap, Department of Plant Sciences
    Project Taken

    Project Description:

    While biodiversity has traditionally been described based on observed species occurrences, species may also be missing from suitable environments. Such species constitute the "dark diversity" of a community, and their absence may be explained by different ecological and environmental factors. Global change, particularly climate change, may increasingly be to blame for species absences, especially those predisposed to dispersal limitation or climate sensitivity due to their traits and evolutionary history. However, dark diversity is difficult to measure systematically and thus to attribute to underlying causes. Using a global network of vegetation monitoring sites DarkDivNet, the overall objective of this project is to determine the role of climate change in explaining why species are absent from sites, drawing on rich methods in comparative phylogenetics.

    There are three main strands of the internship.
    First, you will relate the number of species absences in more than 80 sites worldwide to the velocity of local climate change. You will collate species-level data from the provided site datasets and new datasets of climate velocity from publicly available climate layers to answer whether sites that are experience greater climate change are effectively missing more species from local communities, i.e. greater dark diversity.
    Second, you will test if the resulting climate sensitivity of dark diversity shows phylogenetic structure, indicating the importance of evolutionary history in predicting climate-induced absences. To do this, you will map climate sensitivity rates, as well as species-level contributions to dark diversity across sites, on a publicly available mega-phylogeny of plants.
    Third, if time allows, the student will test the importance of traits in explaining the relationship between dark diversity and climate change sensitivity. Traits could include dispersal limitation, based on seed size and height, plant habit, and climate niche inferred by a physiologically based climate niche model. Teasing apart the factors involved in species exclusion from sites experiencing climate change is crucial to understanding the species-level impact of climate change on biodiversity.

    While all three objectives are valuable, the project is flexible based on the student's interests and practical experience. Completing two of the three objectives would be a worthwhile outcome and contribute to an eventual scientific publication. The project will provide experience with computational skills, data management of diverse sources and formats (survey, spatial, phylogenetic, trait), ecological and phylogenetic statistics, and climate niche modelling. Ideally, you will have some experience with the R programming language and a strong interest in furthering your skills in this area.

  • The student must be comfortable with the R programming language.
  • Full project details and eligibilty criteria are available on the Cambridge NERC Doctoral Training Partnerships website.
  • The duration of REPs will be eight weeks over the summer holiday period (mid-July to September).
  • If the project has the ability to be carried out remotely, it will be noted in the project details link above.
  • Deadline for applications, including one academic reference, must be received by Wednesday 30 June 2021.

  • Insertion Date: 11 June 2021


    (Back to List)

    Transient moulin morphology as a window on glacial rheology

    Supervisor: Dr. Jerome Neufeld, Department of Applied Mathematics and Theoretical Physics
    Project Taken

    Project Description:

    Moulins are vertical conduits (pipes) on the Greenland ice sheet that act as key conduits of meltwater from the surface to the base of the ice sheet, where the seasonal addition of meltwater plays a key role in the summer speed-up and winter slow-down of the ice sheet. The morphology of moulins is transient, and is importantly driven no only by variability to the local surface hydrology and melt rate on hourly to seasonal timescales, but is also sensitive to the local rheology of the ice through creep closure of the conduit. This project will seek to understand the controls on the morphology of moulins, from formation through to closure, by the careful analysis of analogue laboratory experiments on the creep closure of cylindrical cavities in complex rheological fluids exhibiting both power law and yield-strength behaviour. These laboratory studies will be conducted in the GK Batchelor laboratory in DAMTP, and will utilise high-speed imagery of the cross-section of idealised moulins to compare against existing theory for power-law rheology moulins.

    This theory will be extended to examine the collapse of yield-strength materials, with the aim of determining the scaling and geometry of the depth of partial closure for moulins. The results will be compared with published field data on the shallow morphology of moulins, and used to interpret the depth-dependent rheology of glacial ice sheets.

  • Full project details and eligibilty criteria are available on the Cambridge NERC Doctoral Training Partnerships website.
  • The duration of REPs will be eight weeks over the summer holiday period (mid-July to September).
  • If the project has the ability to be carried out remotely, it will be noted in the project details link above.
  • Deadline for applications, including one academic reference, must be received by Wednesday 30 June 2021.

  • Insertion Date: 11 June 2021


    (Back to List)

    They might be giants: Reconstructing body size in relation to climate change in Cenozoic snakes

    Supervisor: Dr. Jason Head, Department of Zoology
    Project Taken

    Project Description:

    Body size is an important functional trait for examining biotic responses to environmental change through deep time and across ecosystems, and reliable size proxies have been developed for many vertebrate clades from the fossil record. Snakes are a hyper-diverse radiation with repeatedly evolution of giant body sizes over the last 66 million years, but precise, reliable body size estimates for fossils have yet to be developed, limiting the ability to examine the relationship between gigantism and climate through time.

    In this project, the student will develop clade-specific linear regression models that predict body size (length and mass) from vertebral measurements based on examination of museum skeletal specimens and segmentation of Computed-Tomographic data sets. The student will develop models for boids (boas, anacondas), pythonids (pythons), and elapoids (cobras and their relatives), and will apply the models to the fossil record in order to reconstruct the tempo and mode of body size evolution in these clades. Body size patterns will be compared with regional and global palaeoclimate proxy data to determine if environmental transitions directly drive body size change through time. Results can be compared with near- and long-term climate model projections to forecast snake faunal responses to future anthropogenic environmental change.

  • The student should have an interest in vertebrate palaeontology and a basic understanding of the fossil record.
  • Full project details and eligibilty criteria are available on the Cambridge NERC Doctoral Training Partnerships website.
  • The duration of REPs will be eight weeks over the summer holiday period (mid-July to September).
  • If the project has the ability to be carried out remotely, it will be noted in the project details link above.
  • Deadline for applications, including one academic reference, must be received by Wednesday 30 June 2021.

  • Insertion Date: 11 June 2021


    (Back to List)

    Carbon storage in Salt Marsh Environments, a machine learning and field based approach

    Supervisor: Dr. Alexandra Turchyn, Department of Earth Sciences
    Project Taken

    Project Description:

    The carbon budget of salt marsh ecosystems has been suggested to be globally significant, although the total carbon stored in highly dynamic salt marsh sediments is currently poorly constrained (1, 2). Anthropogenic climate change and sea level rise is likely to have a significant impact on these ecosystems around the UK and quantifying the current carbon storage will help plan for the future. This research will be based upon our previous studies on the geochemistry of marginal marine sediments (3, 4), and in particular we will build on our previous work using machine learning algorithms to understand the distribution of salt marsh sediment geochemistry. This will be done through applying image recognition and data extraction to predict the sediment type for a range of salt marshes on the Norfolk coast, and work towards building a classification model for UK-wide salt marsh systems. We have a working neural network where this approach has been applied and which classifies the pond sediment geochemistry with over 90% accuracy.

    The second aspect of the project will be to ground truth the carbon content of the sediments. We find that the pond sediment geochemistry is either sulfide rich or iron rich and now that we can use machine learning algorithms to predict and map out the geochemical distribution, we can add the dimension of carbon content and overall carbon storage to our understanding of the system. The summer intern will do field work with the group to collect sediments and combust them to measure the carbon content and use this with the neural network to map out not only the distribution of pond sediment geochemistry but also the carbon storage of the marginal marine ecosystem.

  • Full project details and eligibilty criteria are available on the Cambridge NERC Doctoral Training Partnerships website.
  • The duration of REPs will be eight weeks over the summer holiday period (mid-July to September).
  • If the project has the ability to be carried out remotely, it will be noted in the project details link above.
  • Deadline for applications, including one academic reference, must be received by Wednesday 30 June 2021.

  • Insertion Date: 11 June 2021


    (Back to List)

    Cooperative Machine Learning for Facial Expression Analysis

    Lead Supervisor: Dr. Hatice Gunes of the Department of Computer Science & Technology
    Project Taken

    Project Description:

    Majority of the existing automatic facial expression analysis approaches depend on supervised learning strategies, where the input data is paired with the corresponding human annotation for training the models. However, due to variations in recording conditions and annotation strategies, the trained models usually do not generalize well to real-world scenarios. It is also very time- and effort-consuming to manually provide per-frame annotations for new face videos.

    This project will therefore aim to address this problem by developing a new cooperative learning strategy that allows the model to be trained by a small number of annotations but achieve good generalization capability for face videos recorded under various conditions. Evaluation experiments will be conducted on large-scale facial image datasets.

  • The ideal candidate should have good programming skills in Python. Knowledge and some level of experience with deep learning are also needed.
  • Basic knowledge of computer vision is a plus.
  • The project can be done in-person and remotely, as long as the student has a computer/laptop with GPU.
  • Duration: 10 weeks.

  • If interested please contact the co-supervisor, Siyang Song to register your interest or to apply.

    Insertion Date: 3 March 2021


    (Back to List)

    Learning 3D Printers

    Contact: Lead Supervisor: Dr. Sebastian Pattinson of the Department of Engineering
    Project Taken

     European Union citizens ONLY

    Project Description:

    3D printing is a relatively novel technique where parts are built by the layer-by-layer deposition of material. It offers virtually limitless opportunities to design and manufacture devices because it enables the control of geometry, material composition, and processing conditions at every point in an individual object. There is thus considerable excitement over the potential for 3D printing make better materials and devices, including through the use of complex and locally varying structure. Optimizing across such a large parameter space is difficult, though, and novel learning techniques are likely key to addressing this challenge.

    This project will involve working with the group to build novel learning 3D printing systems and developing their capabilities across different materials. There will be opportunities for hardware, software, and material development and the project will be based at the Institute for Manufacturing, though the specifics depend on the pandemic conditions at the time.

  • Background knowledge of 3D printing, electronics, software development, materials, is useful but not essential.
  • Duration: 8-10 weeks.
  • Location: Institute for Manufacturing in West Cambridge.
  • The project taking place is subject to funding, which is expected to be confirmed no later than March 2021.

  • Please apply to the lead Supervisor, Dr. Sebastian Pattinson as early as possible.

    Insertion Date: 25 February 2021


    (Back to List)

    Optimising a domain-specific language for binary modification

    Contact: Lead Supervisor: Dr. Timothy Jones of the Department of Computer Science & Technology
    Project Taken

    Project Description:

    Within my research group we have been building tools based around binary modification. Our latest is called Cinnamon, a compiler for a domain-specific language for binary instrumentation and profiling, which is built on DynamoRIO, an open-source runtime code manipulation system.

    Currently, Cinnamon's implementation inserts call-backs into the dynamic code stream to perform instrumentation. However, this incurrs a large performance overhead. The aim of this project is to mitigate this by instead inserting assembly code for the call-back function directly into the dynamic stream of instructions, using DynamoRIO's code manipulation interface. A possible extension to this project could be the implementation of instruction manipulation within the DSL, which currently only allows profiling code to be inserted.

  • Essential knowledge, skills and attributes: Good coding skills in C and C++ is essential. Knowledge of compilers and disassemblers would be useful, but is not required.
  • Duration: 10 weeks.
  • This project can be carried on remotely or in person.

  • If interested please contact Dr. Timothy Jones to register your interest or to apply.

    Insertion Date: 25 February 2021


    (Back to List)

    Trace optimisation for DynamoRIO binary instrumentation

    Contact: Lead Supervisor: Dr. Timothy Jones of the Department of Computer Science & Technology
    Project Taken

    Project Description:

    Within my research group we have been building tools based around binary modification. We have created Janus, a binary parallelisation tool, and Cinnamon, a compiler for a domain-specific language for binary instrumentation and profiling. These are built on Janus, an open-source runtime code manipulation system.

    This has excellent support for the x86/amd64 architecture but only basic support for Arm's 64-bit ISA, AArch64. The main missing piece is support for trace optimisation, where hot paths through the code are identified at runtime, instructions along them joined together into a single entity (a trace) and then optimised as a whole. This provides more opportunities for program transformation and a significant performance increase.

    The aim of this project, therefore, is to implement trace optimisations for AArch64 in DynamoRIO. Inspiration can be taken from x86 implementation, but it's hoped that there will be Arm-specific optimisations that can be applied too.

  • Essential knowledge, skills and attributes: Good coding skills in C and C++ is essential. Knowledge of compilers and disassemblers would be useful, but is not required.
  • Duration: 10 weeks.
  • This project can be carried on remotely or in person.

  • If interested please contact Dr. Timothy Jones to register your interest or to apply.

    Insertion Date: 25 February 2021


    (Back to List)

    Deep Learning for shape measurement of biomimetic cell membrane models

    Lead Supervisor: Prof. Ulrich Keyser of the Department of Physics

    Secondary Supervisor: Marcus Fletcher of the Department of Physics

    Project Taken

    Project Description:

    Phospholipid vesicles are an ideal model for a cell's membrane. Several methods exist to produce them synthetically, however their control over membrane biophysical parameters is not well understood [2]. For example, a researcher looking at vesicles prepared for a typical experiment may observe a wide variety of complex morphologies. These shapes may look like simple spheres, dumbbells, necklaces and long tentacle-like tubules. The spontaneous curvature model [1] relates the shape of the vesicle to it's 'in-situ' material parameters and geometric constraints. Thus, measuring shapes leads to membrane biophysical parameters. Accurately determining the shape of large samples of vesicles opens the door to characterising the distribution of biophysical parameters which fix the spontaneous curvature of the membrane. Measuring such shapes is time-consuming and error-prone with manual image analysis. We propose to use machine learning image analysis [3] to detect vesicle shapes more precisely to characterize the distributions of membrane parameters in our vesicle samples.

      [1] Seifert, Udo, Karin Berndl, and Reinhard Lipowsky. "Shape transformations of vesicles: Phase diagram for spontaneous-curvature and bilayer-coupling models." Physical review A 44.2 (1991): 1182.
      [2] Deshpande, Siddharth, et al. "Octanol-assisted liposome assembly on chip." Nature communications 7.1 (2016): 1-9.
      [3] Ren, Shaoqing, et al. "Faster r-cnn: Towards real-time object detection with region proposal networks." Advances in neural information processing systems . 2015

  • The ideal candidate will have introductory Python (or other languages) experience.
  • Experience with Goole Colab would be advantageous.
  • Duration: This project will be run for 10 weeks in the long vacation.
  • This project is fully computational (although can be extended to include lab work if permitted).
  • The student can expect to work completely remotely. We had success running a similar project for a part III student with completely remote supervision.

  • Please send an email with your CV attached to Prof. Ulrich Keyser to register your interest or to apply.

    Insertion Date: 10 March 2021


    (Back to List)

    Advancing image processing of cell microscopy by tightening the hardware-software connection

    Lead Supervisor: Prof. Pietro Cicuta of the Department of Physics
    Project Taken

    Project Description:

    Recognising the boundaries of cells (segmentation) is a key and complex stage of a cell microscopy experiment, crucial to extract data in a systematic form on which to address statistical and biological physics questions. Traditional approaches relied on combinations of image filters.
    In recent years these have been complemented by machine learning and deep learning approaches. Regardless of the "flavor", by and large these methods have been applied on the individual images of a series. In a situation where we have control over the microscope acquisition parameters, there are many things that can be tested: acquisition of the same object in multiple wavelengths; structured illumination (sequence of off-axis source); rapid acquisition of a sequence of frames; modulation of the focal height. These approaches all return a small "stack" to represent one "image" of the sample, i.e. they add one or more dimensions to the data. The project will explore how to make the most of these multidimensional stacks in the context of cell segmentation.

    If access to the lab is possible, the project can have a data acquisition component, on some "simple" to grow biological system, so that the student can control the nature of their dataset. Otherwise, the project will be purely a coding and data analysis project, with images supplied by us. In both cases, a desire to code and to delve into the substantial literature on image analysis is required. Various existing image analysis toolboxes and libraries have to be tested, compared and adapted. Some are coded in python, some in Matlab, some in C++; some work better in Windows others in Linux. An ideal student will have some experience in coding and compiling across these languages and operating systems.

  • Duration: 8-10 weeks- flexible during the out of term period.

  • Please send an email with your CV attached to Prof. Pietro Cicuta to register your interest or to apply.

    Insertion Date: 9 March 2021


    (Back to List)

    Designing an agile quadrotor platform for multi-agent research

    Contact: Lead Supervisor: Dr. Amanda Prorok of the Department of Computer Science & Technology
    Project Taken

    Project Description:

    We are researching multi-agent coordination and would like to demonstrate recent research in real-world experiments. To that end, we need an agile drone platform. Such platforms are not available off-the-shelf, but the FPV (first-person view) flight community has a broad range of components for the hobby market. The goal of this project is to create a research platform consisting of multiple identical quadrotors from scratch, ideally using existing components from the FPV community, that is capable of performing manoeuvres as demonstrated in this video.

    Overview of tasks:

  • Understand and summarize the requirements and propose a quadrotor setup.
  • Create a bill of material and build multiple prototypes iteratively.
  • Assemble and test the quadrotors.
  • Develop a software framework that allows local control of quadrotors and create relevant documentation.
  • Demonstrate the capabilities of the developed platform in a motion capture system.

  • Desired skills:

  • Eager to work with experimental setup (hands-on).
  • Previous demonstrated experience in building robotic platforms, ideally (FPV) quadrotors.
  • Good understanding of basic electrical and mechanical engineering.
  • Good programming skills

  • This project is not suitable for remote working and is dependent upon the Covid-19 restrictions permitting it to be undertaken.
  • Location: Department of Computer Science and Technology in West Cambridge.
  • Duration: 8-10 weeks.
  • This UROP could lead into a final year project on similar topics.

  • If interested please contact Dr. Amanda Prorok to register your interest or to apply.

    Insertion Date: 25 February 2021


    (Back to List)

    Metasurface optics for miniaturised endoscopic imaging

    Lead Supervisor: Dr. Calum Williams of the Department of Physics
    Project Taken

    Project Description:

    Endoscopy enables imaging of tissue deep inside the human body. The continued drive to miniaturise endoscopic imaging technologies-in order to provide imaging to currently 'blind' procedures, and guide future minimally invasive procedures -is limited due to the challenges in shrinking existing sub-system optical components (e.g. lenses, filters, etc.) [1]. These challenges include manufacturability, increased aberrations and narrow field-of-view. Recently, ultrathin optical components (metasurfaces) been developed which instead of refracting the light, utilise the collective scattering of sub-wavelength resonators. Through design, unprecedented control of the optical wavefront can be realised with a single layer [2].

    This project aims to develop metasurface optics specifically designed for miniaturised endoscopic imaging i.e. wide field-of-view achromatic flat lenses. The student will: investigate the state-of-the-art in metasurfaces and requirements for endoscopic imaging; perform electromagnetic simulations of their designs using commercial software (Lumerical FDTD Solutions); optimise designs based on use-case requirements, and (lab access and progress dependent) nanofabricate/characterise these devices in collaboration with group members.


    [1] H. McGoran et al. World J Gastroenterol., 25(30), 4051-4060, 2019.
    [2] D. Lee et al. Nanoscale Adv., 2, 605-625, 2020.

  • The ideal student would have a strong interest in applied physics, optics and imaging.
  • Experience in programming (MATLAB/Python) would be advantageous.
  • They should enjoy working at the interface between different disciplines, and be keen to take the initiative with independent creativity and problem solving.
  • We strongly value a student with the desire to learn, create and innovate.
  • Duration: 10 weeks. Dates can be flexible (proposed: ~5th July-10th Sept.)

  • Please send an email with your CV attached to Dr. Calum Williams to register your interest or to apply.

    Insertion Date: 9 March 2021


    (Back to List)

    Structural testing of optimised concrete shells built using automation and robotics

    Contact: Lead Supervisor: Dr John Orr of the Department of Engineering
    Project Taken

     European Union citizens ONLY

    Project Description:

    The construction industry is responsible for nearly 50% of the UK's carbon emissions, with cement production only causing 7% of global greenhouse gas emissions. Whereas the global population is growing, we need to build more with less.

    The ACORN project on Automating Concrete Construction fuses computational design and digital fabrication to design and build low-carbon concrete shell structures as building slabs in the place of our current highly-reinforced thick plates. The team in the Structures Research Group at the University of Cambridge focuses on the use of reconfigurable actuated moulds, robotic concrete spraying and filament winding to produce these advanced structural forms and patterns.

    We are looking for an undergraduate student to contribute to our project through structural testing of our concrete shells to understand their behaviour and performance. The student will also assist the researchers and technicians with prototyping and automation of the fabrication process, aiming at the building of a large-scale demonstrator at the end of 2021. Experience in structural analysis and testing is important, whereas simple prototyping and basic coding are an advantage.

  • This project will be co-supervised by Dr Robin Oval of the Department of Engineering.
  • Duration: Ideally 12 weeks around the summer.
  • Location: The Civil Engineering Building in West Cambridge.
  • Continuation as a 4th-year project on this topic or another aspect of the ACORN project is possible.
  • The project taking place is subject to funding, which is expected to be confirmed no later than March 2021.

  • Please apply to the lead Supervisor, Dr John Orr.

    Insertion Date: 21 January 2021


    (Back to List)

    The molecular mechanisms controlling wrinkle formation in B. subtilis biofilms

    Lead Supervisor: Dr. Diana Fusco of the Cavendish Laboratory, Department of Physics
    Project Taken

    Project Description:

    Biofilms are self-assembled communities of bacteria. Bacillus subtilis is a gram+ bacterium that is able to form robust biofilms known for their wrinkly architecture. While it is well documented that within this biofilm cells display different behaviour, the coupling between this heterogeneity and the wrinkling architecture is still unclear. More specifically, what types of cell behaviours are responsible for the wrinkled structure and how does the biofilm morphology feedback on cell behaviour? To answer this question, various promoters of genes involved in different aspect of the bacterial physiology will be cloned in transcriptional fusion with reporter genes in order to visualize the heterogeneity in gene expression within a biofilm and the wrinkling patterns. Biofilms grown using strains containing the transcriptional fusions will be monitored using epifluorescence microscopy.

    During the project, the student will gain experience in experimental design, molecular biology techniques, microbiology culture techniques, and image analysis.

    The day-to-day supervisor is Dr. Racha Majed.

  • Interest in biological physics and molecular biology is required.
  • Experience in wet-lab techniques and/or image analysis is highly welcome.
  • This opportunity is open only to students that have at least one year to study before graduation.
  • The duration of the project is 10 weeks, with a preferable start date in June. The main location of the work is the Physics of Medicine Building at the Cavendish.
  • The project is mainly experimental, so it cannot be carried out remotely.

  • Please send an email with your CV attached to Dr. Diana Fusco as early as possible to apply.

    Insertion Date: 9 March 2021


    (Back to List)

    Differential Privacy

    Lead Supervisor: Dr. Nicolas Lane of the Department of Computer Science & Technology
    Co-supervisor: Dr. Pedro Porto Buarque de Gusmao of the Department of Computer Science & Technology
    Project Taken

    Project Description:

    Federated learning offers increased data privacy as only the model's weights, and not the underlying data, are shared with a server during training. However, recent results in literature show that it is still possible to obtain private data when these shared weights are leaked. This technical liability calls for an additional layer to improve user's privacy.

    In this project, you will implement a Flower example that demonstrated how Differential Privacy can be used on the client. Flower is an open-source framework for performing Federated Learning research and teaching with many contributors in the Cambridge Computer Lab. Flower has been used for teaching this year. This project would be done by a single student but conducted within a vibrant group of other researchers that are studying other aspects of federated learning within the Machine Learning Systems lab.

    Expected Outcome:

  • Create a code example that implements Differential Privacy in either Tensorflow or PyTorch.
  • Test and document the new example.
  • Write a blog post describing the new code example.

  • The ideal candidate should have some understanding of security principles (encryption/decryption);
  • A basic understanding of differential privacy.; and
  • A basic understanding of machine learning and deep learning (TensorFlow or PyTorch).
  • The project can be carried out remotely.
  • Duration: 10 weeks.
  • If you have applied for the Secure Aggregation position, you are also being considered for this position as well.

  • If interested please contact Dr. Nicolas Lane or Dr. Pedro Porto Buarque de Gusmao to register your interest or to apply.

    Insertion Date: 14 April 2021


    (Back to List)

    Modelling biological tissues under loading

    Lead Supervisor: Dr. Thierry Savin of the Department of Engineering
    Project Taken

    Project Description:

    This project applies engineering computing methods to study biological tissues under load. In many biological processes, such as embryogenesis, wound healing or cancer, soft tissues must grow while being stretched. To provide a fundamental understanding of the mechanical feedback on growth, we developed a simple model of tissues, based on a network of elastic fibres inspired by the collagenous microstructure of the extracellular matrix. The aim of the project is to modify the model to include rearrangement and fibre growth, and compare the results with experimental data.

  • The ideal student would have a strong interest in bioengineering, and have experience in Python programming and some knowledge of Abaqus finite element software.
  • The student will closely work with a PhD student.
  • The estimated duration of the project is 8 to 10 weeks, during the summer.
  • This project can be continued into a 4th year project that will build on the findings.
  • This project is currently seeking funding, which is yet to be confirmed.

  • If interested, please contact Dr. Thierry Savin with your CV attached.

    Insertion Date: 12 April 2021


    (Back to List)

    Secure Aggregation in Federated Learning

    Lead Supervisor: Dr. Nicolas Lane of the Department of Computer Science & Technology
    Co-supervisor: Dr. Pedro Porto Buarque de Gusmao of the Department of Computer Science & Technology
    Project Taken

    Project Description:

    Federated Learning is a privacy-oriented method for training neural networks that relies on thousands of connected devices to collaboratively learn a shared model without sharing any data. In this process, each connected device iteratively trains the model on their local data and sends the updated model to a central server thereafter. The server, in turn, aggregates all updates and produces an improved version of the shared model. Secure Aggregation protects these model updates from being individually analysed by giving the server access only to the aggregated model. This prevents the server from being able to "peek" into the model update from a single device and learn sensitive information.

    In this project, you will implement a Secure Aggregation method for the Flower federated learning framework and create a code example that demonstrates the usage of Secure Aggregation with Flower. Flower is an open-source framework for performing Federated Learning research and teaching with many contributors in the Cambridge Computer Lab. Flower has been used for teaching this year. This project would be done by a single student but conducted within a vibrant group of other researchers that are studying other aspects of federated learning within the Machine Learning Systems lab.

    Expected Outcome:

  • Implementation definition for Secure Aggregation based on state of the art approaches.
  • Implement changes in the current message protocol (i.e. messages exchanged between server and clients) to allow for Secure Aggregation.
  • Implement the required server-side and client-side logic.
  • Create a code example using the new functionality.
  • Documentation describing how to use the new feature.
  • Optional: a blog post about the new feature.

  • The ideal candidate should have knowledge of python programming;
  • Some understanding of security principles (encryption/decryption); and
  • A basic understanding of machine learning.
  • The project can be carried out remotely.
  • Duration: 10 weeks.

  • If interested please contact Dr. Nicolas Lane or Dr. Pedro Porto Buarque de Gusmao to register your interest or to apply.

    Insertion Date: 19 March 2021


    (Back to List)

    Bayesian machine learning and theory in cosmology and particle physics

    Lead Supervisor: Dr. Will Handley, Cavendish Laboratory Astrophysics Group (Kavli Institute for Cosmology), Department of Physics
    Project Taken

    Project Description:

    In this project the student will work with Dr. Will Handley and his team investigating the development and application of Bayesian machine learning techniques to modern and future cosmological and particle physics datasets. The precise details of the project will be tailored to the student interest and skill set, but possible topics/projects include:

    1. Developing machine learning algorithms for nested sampling and applying these to cosmological data sets
      - PolyChord: next-generation nested sampling
      - Quantifying the global parameter tensions between ACT, SPT and Planck
    2. Model independent reconstruction of the primordial universe from cosmic microwave background data and cosmic dawn data
      - Bayesian inflationary reconstructions from Planck 2018 data
    3. Developing and applying mathematical schemes for disentangling physical signatures in the primordial universe
      - Primordial power spectra for curved inflating universes
      - Analytical approximations for curved primordial power spectra
    4. Combining particle physics and cosmological data as part of the GAMBIT team
      - The Global And Modular BSM Inference Tool
      - The Global And Modular BSM Inference ToolCosmoBit: A GAMBIT module for computing cosmological observables and likelihoods
      - Strengthening the bound on the mass of the lightest neutrino with terrestrial and cosmological experiments
    A list of Dr. Handley's articles can be found here.

    Over the course of the project students can expect to learn some/all of:
    - up-to-date cosmological research questions
    - Science grade python
    - High performance computing
    - Bayesian inference
    - Machine learning
    - Computer algebra

  • Essential: Three years of undergraduate physics or equivalent, basic to intermediate Python experience and good programming skills, strong mathematical skills.
  • Desirable: Interest/knowledge of general relativity/cosmology and experience using Mathematica/Maple/Computer algebra.
  • Duration: 10 weeks.
  • This project could lead into a final year project on similar topics.

  • Please send an email with your CV attached to Dr. Will Handley to register your interest or to apply.

    Insertion Date: 16 March 2021


    (Back to List)

    AI-empowered Functional Genomics

    Lead Supervisor: Prof. Anna Korhonen, Department of Theoretical & Applied Linguistics
    Project Taken

    Project Description:

    Language Technology Laboratory

    We are developing an open source AI tool that can automate functional genomics. In particular, the tool will automatically identify Essential Genes (EGs) which can be important as targets of treatment (e.g. for cancer) or sources of side effects of drugs. Current EG identification requires costly and time-consuming wet lab experiments that involve gene manipulation. Our novel AI tool will use advanced natural language processing and machine learning to text mine all publicly available scientific articles and experiment records to construct a dynamic knowledge base that biomedical scientists can search and interact with using an online system. When adopted by biomedical scientists, the tool can reduce the costs of unnecessary lab experiments, significantly cut the time spent on literature survey, and catalyse drug discovery.

    The goal of this project is to develop a named entity tagger and a text sequence classifier for extracting key information about experimental setups as described in published scientific articles downloaded from PubMed. These include identification of perturbing actions applied on genes and proteins, identifying the effects of these actions as well as other context. Prior knowledge or experience in machine learning (or natural language processing) is an advantage.

  • Essential: Python 3 and basic Linux/Unix shell skills.
  • Desirable: Deep Learning packages such as Pytorch or Tensor Flow and NLP packages (e.g. NLTK, SpaCy).
  • Duration: 10 weeks.
  • This project can be carried on remotely if needed.

  • Please send an email with your CV attached to Prof. Anna Korhonen to register your interest or to apply.

    Insertion Date: 16 March 2021


    (Back to List)

    Automated Literature-Based Discovery for Cancer Research

    Lead Supervisor: Prof. Anna Korhonen, Department of Theoretical & Applied Linguistics
    Project Taken

    Project Description:

    Language Technology Laboratory

    The enormous size of the scientific literature makes it increasingly difficult for biomedical researchers to stay up to date on all developments in their field. To address this problem, we have developed LION LBD - an open source system that enables automated literature-based discovery. Using this system, scientists can navigate published scientific literature and automatically generate hypotheses for their research in molecular biology or cancer. The system is based on state-of-the-art methods for natural language processing and machine learning and offers both an interactive web-based interface for users and a programmable API.

    The goal of this project is to upgrade some of the classifiers used in this project to new generation of classifiers (e.g. Transformers), and/or to build an auto-update feature that checks for latest updates released on PubMed, and PubTator.

  • Essential: Python 3 and solid Linux/Unix shell skills.
  • Desirable: SQL (PostgreSQL database) and deep learning packages such as Pytorch or Tensor Flow.
  • Duration: 10 weeks.
  • This project can be carried on remotely if needed.

  • Please send an email with your CV attached to Prof. Anna Korhonen to register your interest or to apply.

    Insertion Date: 16 March 2021


    (Back to List)

    Greenhouse gas removal evaluation

    Lead Supervisor: Dr. Shaun Fitzgerald of the Centre for Climate Repair at Cambridge
    Co-supervisor: Dr. Hugh Hunt of the Department of Engineering
    Project Taken

    Project Description:

    There are wide range of options for removing greenhouse gases from the atmosphere, which range from nature-based approaches such as afforestation through to engineering-led solutions involving direct air capture of carbon dioxide and storage of the gas in old oil and gas reservoirs.
    This project will involve the development of a framework which can be used to help compare different approaches in order to assess options. The factors which need to included will be wide-ranging and will likely for example: development costs and likely timescales of research/development; scalability; impact on ecosystems; use of resource such as land, water, energy and materials; social acceptability; social justice.
    The project will involve drawing together relevant literature for greenhouse gas removal approaches, interviewing academic experts, interviews with those from industry who have interests in particular approaches, and identifying any existing evaluation frameworks and building upon these. The project will then seek to use the framework, and start building a database of projects and applications so that they can be compared.
    For more about the Centre for Climate Repair, do visit our website Centre for Climate Repair.

  • The project will last for 8-10 weeks. The student will be able to start in early July and can finish as late as end of October.
  • This project can be carried out remotely.
  • Students from all departments are able to apply, although it will be preferable for them to have at least one A level in mathematics, physics or chemistry.

  • Please send a cover email, mentioning which specific project you are applying for, with your CV attached by 5pm on Thursday 22 April 2021 to Katie Parker.
    There will be a webinar on Friday 23 April 2-3pm for you to find out more (invite will follow when you apply).

    Insertion Date: 25 March 2021
    Edited:12 April 2021


    (Back to List)

    Ice Thickening

    Lead Supervisor: Dr. Shaun Fitzgerald of the Centre for Climate Repair at Cambridge
    Project Taken

    Project Description:

    Increasing loss of ice in the Arctic is a serious problem and leading to reduced albedo. Consequently, more solar energy is absorbed by the ocean system and a positive feedback loop for climate change is being established. There are concerns that the pace of transition to net zero will not be sufficiently fast to protect the ecosystems in the Arctic.
    One of the concepts which has been suggested involves increasing the thickness of ice formed during the Arctic winter. The concepts have involved spraying seawater onto the top surface of the sea ice in order to expose more water to the cold Arctic temperatures, and breaking up the newly formed sea ice in early winter in order to increase the rate of freezing. Ostensibly these approaches also involve increasing the thickness of the sea ice which forms. If sea ice develops to be more than ~1m thick then it will likely last a whole season. This project will involve a review of the existing literature regarding such techniques, and then modelling of the formation of sea ice under such proposed conditions.
    For more about the Centre for Climate Repair, do visit our website Centre for Climate Repair.

  • The project will last for 8-10 weeks. The student will be able to start in early July and can finish as late as end of October.
  • This project can be carried out remotely.
  • Students from engineering, applied mathematics, and physical sciences are best suited to this project.
  • This project may provide an opportunity for continuation into a 4th year project.

  • Please send a cover email, mentioning which specific project you are applying for, with your CV attached by 5pm on Thursday 22 April 2021 to Katie Parker.
    There will be a webinar on Friday 23 April 2-3pm for you to find out more (invite will follow when you apply).

    Insertion Date: 25 March 2021
    Edited:12 April 2021


    (Back to List)

    Visualisation for Bayesian Decision Making

    Lead Supervisor: Prof. Alan Blackwell of the Department of Computer Science & Technology
    Project Taken

    Project Description:

    The goal of this project is to create a prototype of a new interactive visualisation, using Bayesian information flows as an intelligent tool that helps untrained users to interact with autonomous agents. The technical approach will be related to the design of probabilistic programming languages, including modelling of statistical distributions, as applied in machine learning and data science. The goal here will be to create new interaction modes for coordinating autonomous systems through intuitive graphical programming. The project will involve several different programming languages and tools, and candidates should have reasonably broad programming experience, as well as an interest in visualisation and/or Bayesian statistics. There will be an opportunity to work with others in a team, including graduate students carrying out specialist research on this topic for audiences including medical decision making and environmental science in Africa, as well as with industry researchers who are sponsoring the project from Boeing Research and Technology.

  • The ideal candidate must have broad programming experience.
  • It is desirable for you to have an interest in visualisation and/or Bayesian statistics.
  • The duration is for a maximum of 10 weeks, can be less if needed.
  • The project can be done remotely or in person.

  • If interested please contact Prof. Alan Blackwell to register your interest or to apply.

    Insertion Date: 14 April 2021


    (Back to List)

    Food handling robot development and experiment

    Lead Supervisor: Dr. Fumiya Iida of the Department of Engineering
    Project Taken

    Project Description:

    This project aims to develop a food handling robot to be used in packing and quality control process in food logistics companies. We use a commercially available dual-arm robot manipulator and customise the platform for our specific target applications for our end-user company.

  • You will be expected to programme depth vision as well as motor control architecture by using ROS, together with the other collaborating engineers.
  • The project is planned for 10 weeks and it will be preferable to be carried out in person, but it might be possible to complete the project remotely.

  • Students should submit their cover email and indicate their project preferences, with their CV attached to Dr. Fumiya Iida.
    Applications will be reviewed as they arrive, so an early submission is encouraged.

    Insertion Date: 4 May 2021


    (Back to List)

    Mobile manipulator robot development

    Lead Supervisor: Dr. Fumiya Iida of the Department of Engineering
    Project Taken

    Project Description:

    This project aims to develop a mobile manipulator robot technology applied to elderly assistance applications. By extending the kitchen lab set up we have co-developed so far, we will develop a few demonstration scenarios in which robotic technologies and sensors can be helpful for independent living of elderly individual at home.
    In this project we proposed the development of a mobile robot platform that can be used not only for floor cleaning but also extra functions such as remote monitoring, emergency alerting, simple goods transportation, etc. that are challenging for many elderly for independent living. A mobile robot platform was installed in the kitchen lab, and equipped with sensing and communication devices for such application scenarios.
    With the 10-week UROP project, we will develop the hardware and software components of the platform. The project will also be documented into a technical publication possibly submitted to one of the technical conferences, along with a set of recorded and live demonstrations for the discussions of future commercial applications.

  • You will be expected to programme the robot platform in Python, and have some basic skills of robot hardware development.
  • The project is planned for 10 weeks and it will be preferable to be carried out in person, but it might be possible to complete the project remotely.

  • Students should submit their cover email and indicate their project preferences, with their CV attached to Dr. Fumiya Iida.
    Applications will be reviewed as they arrive, so an early submission is encouraged.

    Insertion Date: 4 May 2021


    (Back to List)

    Source Brightness Measurements for Helium Atom Microscopy

    Lead Supervisor: Dr. Andrew Jardine of the Department of Physics
    Project Taken

    Project Description:

    Scanning helium microscopy (SHeM) is an exciting new form of microscopy for imaging delicate surfaces, and is the subject of a £1M research programme within the SMF group at the Cavendish Laboratory. The brightness of the helium source plays a crucial role in determining image quality, and is determined by a variety of thermodynamic and (both classical and quantum) scattering processes around the supersonic helium expansion. The aim of this project is to experimentally characterise the brightness of the virtual helium source for the new microscope, under a variety of different pressure and flow conditions. The data will be compared to existing models, and will be used to determine the operating conditions in all future instruments. The project will involve hands on laboratory work and subsequent analysis.

  • Tripos experience in a scientific or engineering discipline, and an interest experimental physics research is essential.
  • Experience of at least Part IB Physics, or equivalent, is highly desirable.
  • 10 weeks: There are no timing restrictions over the long vacation period, although the particular dates will need to be agreed before the summer.

  • Please send an email with your CV attached to Dr. Andrew Jardine to register your interest or to apply.

    Insertion Date: 16 March 2021


    (Back to List)

    Structured back gates on GaAs

    Lead Supervisor: Professor Dave Ritchie of the Department of Physics
    Co-supervisor: Dr Chong Chen of the Department of Physics
    Project Taken

    Project Description:

    This project involves a few iterations of experiments designed to understand how to make reliable, reproducible and structured back gates for Molecular beam epitaxy (MBE) grown nano-devices. With this technique, sophisticated devices with a top gate and a back gate are possible. This will enable wave function modulation in the quantum well to either side, or to isolate a particular conducting layer out of 2 or 3 parallel conducting layers without the need for complicated processing. The student will (i) be trained in the design of optical masks using AutoCAD, (ii) study surface morphology using an Atomic Force Microscope and (iii) study surface contamination using Secondary Ion Mass spectroscopy.
    Further reading on the subject can be found here and here.

    The ideal candidate will be:
    1. Highly motivated and enthusiastic
    2. Ability to carry out systematic scientific investigation under guidance and to work independently when possible
    3. Ability to communicate clearly and present data to others
    4. Ability to make clear plan and stick with timelines
    5. A high level of accuracy and attention to detail
    6. Experience with programming in C, C++, Matlab, would be an advantage

  • The project could be organised flexibly based on progress and experiment arrangement as samples will have to be send away for ion implantation and we expect waiting time in between.
  • This project could lead into a final year project on similar topics.

  • Please send an email with your CV attached to Professor Dave Ritchie of the Department of Physics to register your interest or to apply.

    Insertion Date: 25 March 2021


    (Back to List)

    Prospecting for unconventional superconductors

    Lead Supervisor: Prof. Malte Grosche of the Cavendish Laboratory, Department of Physics
    Co-supervisor: Dr. Jiasheng Chen of the Cavendish Laboratory, Department of Physics
    Project Taken

    Project Description:

    Conventional superconductivity, as found in many elements and alloys, tends to be fairly insensitive to pressure, composition or disorder. There are, however, an increasing number of materials in which superconductivity is limited to high quality crystals and to small portions of the pressure-temperature or composition-temperature phase diagram. This suggests that a degree of fine tuning of the effective interaction between the electrons is involved. These *unconventional* superconductors are often found on the threshold of magnetic order, where fluctuations of the local magnetisation can play an important role in binding the Cooper pairs together. The high temperature superconducting cuprates and iron pnictides/chalcogenides fall into this category.

    Unconventional superconductivity is important, because a strong effective pairing interaction will be needed to achieve superconductivity at room temperature, with all the technological benefits this would bring. This may no longer be a fancy dream. Very recently, *conventional* superconductivity was discovered near room temperature in the superhydride LaH10 [1], which however had to be compressed to megabar pressures in order to achieve the high vibration frequencies that are needed for a strong pairing interaction in a conventional superconductor. This demonstrates (i) there's no fundamental reason why superconductivity should be limited to very low temperatures, but also (ii) for it to occur at ambient pressure as well as high temperatures will require a strong non-phonon pairing interaction, as is already manifested in some iron-based and cuprate superconductors. In short, we need new unconventional superconductors.

    In this project, we will use theoretical modelling and numerical calculations to try out guiding principles that may accelerate the search for new unconventional superconductors with a magnetic pairing interaction. This involves (i) modelling of the magnetic fluctuation spectrum in materials near the threshold of magnetic order, (ii) modelling of key elements of the electronic structure, (iii) estimates of superconducting ordering temperatures, similar to the approach shown in [2]. All three steps can be taken from semi-heuristic approximate treatments to arbitrary levels of sophistication. Time permitting, the project can also involve some crystal growth and low temperature/high magnetic field characterisation of candidate materials explored in the theoretical part of the project.

    [1] Drozdov et al., Nature 569, 528 (2019).
    [2] Monthoux & Lonzarich, Physical Review B 59, 14598-14605 (1999).

  • The ideal candidate will have independence, resourcefulness, good communication and teamwork skills.
  • We expect the project to run for a minimum of six and a maximum of ten weeks, distributed over the time mid-June till end-September, depending on the other constraints on the student.

  • Please send an email with your CV attached to Prof. Malte Grosche to register your interest or to apply.

    Insertion Date: 13 April 2021


    (Back to List)

    Monitoring changes in tropical forest carbon storage at the individual tree level

    Lead Supervisor: Dr Toby Jackson of the Department of Plant Sciences / Cambridge Conservation Initiative
    Project Taken

    Project Description:

    Tropical forests mitigate climate change by absorbing some of the carbon we emit, but how is this changing? The most accurate answers to this question come from combining remote sensing (i.e. satellite or airborne imagery) with detailed field data. However, there is an important difference between what the fieldworker sees (trees) and what the camera sees (pixels).

    To bridge this gap, algorithms have been developed which detect individual trees in remote sensing data, but they perform poorly in tropical forests. Most current approaches are hand-tuned and do not leverage recent advances in machine learning methods. However, DeepForest (see paper) substantially improved tree detection accuracy for temperate sites across the US by deploying state-of-the-art machine learning models (such as convolutional neural networks). We will apply this method to tropical forests using high resolution airborne LiDAR data and RGB imagery covering >100,000 trees in protected tropical forests in Sabah, Malaysia in both 2014 and 2020.

    Background introductory reading:
    Coomes et al 2017. Remote Sensing of Environment. Area-based vs tree-centric approaches to mapping forest carbon in Southeast Asian forests from airborne laser scanning data
    Weinstein et al 2021. E-life. A remote sensing derived data set of 100 million individual tree crowns for the National Ecological Observatory Network.

    Brief description of what the student will do, the skills they will gain and the outcome expected from the project:
    You will help apply an existing machine learning algorithm (DeepForest) to tropical forests in Malaysia.
    This will involve:

  • Manually delineating tree crowns in remote sensing images (this is the train / test data)
  • Updating the DeepForest model with this data and assessing its accuracy on the remaining tree crowns
  • Once individual trees have been extracted, calculate their biomass in 2014 and 2020 (around half of a tree’s woody biomass is carbon).

  • The outcome will be an estimate of the change in carbon stored in the tropical forests of Sabah, Malaysia at the individual tree level.
    Skills you will gain:
  • Machine learning in python (you can decide how involved to get)
  • Remote sensing analysis (airborne LiDAR and RGB imagery)
  • Estimating carbon stocks of forests
  • Handling spatial data (R & QGIS)

  • Please read the further particulars about the role and eligibility.

    Required academic background/skills of student:
  • Some familiarity with programming (R or python) and with machine learning concepts is a benefit.
  • However, the main requirement is that you are enthusiastic and diligent.
  • You will be working alongside a postdoc and masters student and we can discuss how much you know / need to learn at the start of the project.


  • The duration is 8 weeks, any time in June-September period.
  • The project will be remote.

  • Students should submit their application by email to the CDT administrator no later than Monday 10 May 2021.

    Insertion Date: 21 April 2021


    (Back to List)

    Assessing the effects of forestry on natural climate solutions

    Lead Supervisor: Dr Andrew Tanentzap of the Department of Plant Sciences
    and Dr Jeremy Fonvielle of the Conservation Research Institute
    Project Taken

    Project Description:

    The forestry sector will likely play a major role in climate change mitigation and the removal of atmospheric CO2 by enabling a shift to wood building materials and producing wood-derived biochar and biofuels. However, these strategies will necessitate an increase in the total production of forest products. Given that global climate commitments rely on promoting carbon sequestration, the role of the forestry sector must be examined carefully. One under-studied component is the connection between terrestrial and aquatic ecosystems. Soil organic matter (OM) is the largest terrestrial reservoir of C. Therefore, if OM-rich soils are increasingly disturbed by mechanized harvest, or become better connected to water flow paths because of forestry practices, there is a heightened potential for terrestrial C to leak into downstream waters and offset C sequestration on land.

    Background introductory reading:
    Kreutzweiser, D. P., P. W. Hazlett, and J. M. Gunn. 2008. Logging impacts on the biogeochemistry of boreal forest soils and nutrient export to aquatic systems: A review. Environmental Reviews 16:157–179.
    Schelker, J., K. Eklof, K. Bishop, and H. Laudon. 2012. Effects of forestry operations on dissolved organic carbon concentrations and export in boreal first-order streams. Journal of Geophysical Research: Biogeosciences 117.

    Brief description of what the student will do, the skills they will gain and the outcome expected from the project:
    This project tests how the composition and flux of OM from streams in Canada's vast boreal forest varies with forest harvest practices. The student will work with a dataset of the optical properties and concentration of dissolved OM in 200 streams and relate these to forest harvest history, land use characteristics, landscape geomorphology, and bioclimatic variables.
    The student will initially spend 4 weeks training, tuning, and validating machine learning models (e.g. random forests) in R or Python using data from the field survey. They will then spend 2 weeks using geospatial tools to extrapolate their models alongside existing rates of forest carbon sequestration to generate a regional carbon budget. The final 2 weeks will be spent writing up the results.
    The desired outcome will be a contribution to a peer-reviewed publication on the unitended consequences of forestry for climate targets. The student will gain competence in machine learning, forest ecology, and biogeochemistry.


    Please read the further particulars about the role and eligibility.

    Required academic background/skills of student:
  • Experience in a programming language, ideally R or Python, is necessary.
  • The duration is 8 weeks, any time in June-September period.
  • The project will be remote.

  • Students should submit their application by email to the CDT administrator no later than Monday 10 May 2021.

    Insertion Date: 21 April 2021


    (Back to List)

    Can Federated Learning Save the Planet?

    Supervisors: Dr. Nicolas Lane of the Department of Computer Science & Technology
    and Dr. Pedro Porto Buarque de Gusmao of the Department of Computer Science & Technology
    Project Taken

    Project Description:

    For the better part of a decade, Deep Learning has been defining the state of the art in various machine learning tasks. Ranging from Computer Vision [1] to Natural Language Processing [2], models have achieved remarkable results at the cost of doubling computation resources every 3.4 months [3]. As hardware efficiency struggles to keep up with the exponential growth in architecture complexity, a similar trend in energy consumption is created as more hardware and cooling systems are required to train modern models. In fact, recent studies [4][5] have shown that training large models in conventional data centres, a sector that already amounts for 0.3% of the world’s carbon emission [6], can cause significant increase in CO2eq production.

    Fortunately, not all hope is lost and a more carbon-friendly way to train neural networks exists. In Federated Learning (FL), training is performed not inside large data centres, but distributed over thousands of mobile devices, such as smartphones, where data is usually collected by the end-users themselves.

    An example of an application currently using FL is the next-word prediction in mobile phones [7]. In this application, each smartphone (client) trains a local network (model) to predict which word the user will type next based on their previous text messages. Trained local models are then sent to a server to perform a much simpler task called aggregation in which a final model will be generated and sent back to all users.

    Besides the privacy-related gains of not having to send user-data to a centralised server, in our recent work [8], [9] we show that FL can also have a positive impact in reducing carbon emissions derived from Deep Learning. Although mobile devices are much less powerful than server GPUs, FL benefits from not needing any cooling and from having wide pools of devices for training.

    This project aims to expand and build on our on-going research into the carbon footprint of federated learning, and how these methods can be made even more environmentally friendly.

    Background introductory reading:
    CO2 in Federated Learning. A first look into the carbon footprint of federated learning. Can Federated Learning Save The Planet?. Quantifying the Carbon Emissions of Machine Learning. How to stop data centres from gobbling up the world's electricity.

    Brief description of what the student will do, the skills they will gain and the outcome expected from the project:
    We have already conducted early-stage research into the carbon overhead of federated learning. This is important because federated learning is an up-and-coming method which is increasingly proliferating, thus we need to carefully understand its impact in the environmental domain. By doing so we can better shape its evolution towards more environmentally friendly designs - an important step that never occurred during the progression and evolution of datacentre based machine learning.
    Future directions we wish to expand into include:

  • Improving the fidelity of our environmental model to better judge the positive/negative factors to carbon pollution due to federated learning.
  • Study typical large scale federated learning deployments, currently our research is fairly limited to standard datasets – but analysis has not been done on the type of large-scale federated learning deployments that are currently run by a number of corporations.
  • Federated learning is likely to be paired with various forms of unsupervised learning and hybrid setups that match label and unlabelled data (along with hybrid centralized and de-centralized forms learning approaches). We wish to perform speculative analysis on what the carbon (and more broadly efficiency) will be of such systems.
  • The work of the above will be a mixture of:

  • Modelling of environmental factors at an improved resolution than we have done so far and likely require integrating concepts from environmental prediction disciplines
  • Extrapolating from early-stage code and research results from the area of self-supervised / unsupervised learning along with federated learning
  • Theoretical analysis where possible, but more often prototypes and experiment-based analysis.

  • Outcomes and benefits to students involved: Working with an active research team and likely producing a workshop paper or more from their efforts.

    Please read the further particulars about the role and eligibility.

    Required academic background/skills of student:
  • Comfort with python and pytorch (ideally).
  • Prior experience with deep learning would be a strong plus.
  • A perfect candidate would also have some experience in federated learning or non-trivial use of deep learning (such as experience in training a large model using large datasets like ImageNet).


  • The duration is 6-8 weeks.
  • The project will be remote.

  • Students should submit their application by email to the CDT administrator no later than Monday 10 May 2021.

    Insertion Date: 21 April 2021


    (Back to List)

    Using machine learning to explore options for replacing lead in lead perovskite solar cells

    Lead Supervisor: Bingqing Cheng of the Department of Computer Science and Technology
    Project Taken

    Project Description:

    Cesium lead halide perovskite (CsPbX3, X = Cl, Br, I) is extremely appealing for making highly efficient solar cells. Unfortunately, the lead element is harmful for the environment as well as the human body. In order to find a substitute for the lead in CsPbX3, we want to model its material properties, and then predict how a substitution of lead by other elements affects the performance.

    Background introductory reading:
    Volker L Deringer, Miguel A Caro, and Gabor Csanyi. Machine Learning Interatomic Potentials as Emerg-ing Tools for Materials Science. In: Advanced Materials 31.46 (2019), p. 1902765.DOI:10.1002/adma.201902765

    Brief description of what the student will do, the skills they will gain and the outcome expected from the project:
    The student will learn about machine learning (ML) potentials, which are ML models that learn atomic interactions from quantum mechanical calculations.
    The student will try to fit a ML potential for CsPbX3 and/or other perovskites.
    The student will get hands-on experience in Python, ML, and running atomistic simulations, as well as gain knowledge on quantum mechanical calculations.


    Please read the further particulars about the role and eligibility.

    Required academic background/skills of student:
  • Python, bash script, familiarity to Linux environments.


  • The duration is 8 weeks.
  • The project will be remote.

  • Students should submit their application by email to the CDT administrator no later than Monday 10 May 2021.

    Insertion Date: 21 April 2021


    (Back to List)

    Can machine learning link ocean data with the polar vortex?
    Using unsupervised classification of ocean data to track changes in Southern Hemisphere winds.

    Supervisors: Emma Boland and Dan Jones of the British Antarctic Survey
    Project Taken

    Project Description:

    The Southern Annular Mode (SAM) describes north-south shifts in the position and strength of the powerful westerly winds that encircle Antarctica, a structure known as the polar vortex. Over recent decades, ozone depletion and greenhouse gas forcing have driven a shift in these winds, leading to warmer and drier conditions in the Southern Hemisphere. Because of its importance in driving heat, rainfall, and fire risks, measuring and predicting changes in the SAM remains an important goal for risk-aware climate science.

    At present, researchers typically use a simple atmospheric pressure difference between two specific latitudes to define the SAM. This straightforward approach may miss the development of regional structures and may not be appropriate in future climates, due to shifts in the strongest winds. Researchers now have an opportunity to develop alternative methods for tracking the SAM, following on from recent successes in applying unsupervised classification to ocean data.

    Background introductory reading:
    Thompson, D.W.J., & Solomon, S. (2002). Interpretation of Recent Southern Hemisphere Climate Change, Science, 895-899.
    Jones, D.C., Holt, H. J., Meijers, A. J. S., & Shuckburgh, E. (2019). Unsupervised clustering of Southern Ocean Argo float temperature profiles, Journal of Geophysical Research: Oceans, 124, 390-402.

    Brief description of what the student will do, the skills they will gain and the outcome expected from the project:
    First, we propose that the student reproduce the results of Jones et al. (2019) which used Gaussian Mixture Modelling (GMM) to cluster temperature profile types. This exercise will give the student an opportunity to get familiar with using GMM and with oceanographic data.
    Next, we propose to train a GMM on ocean temperature data from a UK climate model, which will provide a larger dataset which covers space and time uniformly. The student will then look for correlations between the properties of these classes and the SAM index or proxies that are often used in its place (e.g. atmospheric jet latitude and strength). This will allow the student to develop their analytic and data presentation skills.

    The expected outcomes are:
  • A description of successful GMM clustering of the climate model data.
  • Descriptions of the derived GMM class properties.
  • Descriptions of the statistical relationships between these class properties and SAM properties.

  • Please read the further particulars about the role and eligibility.

    Required academic background/skills of student:
  • Experience with computer code (python, Matlab or similar).
  • Basic knowledge of machine learning techniques (experience applying these techniques is not necessary).
  • Basic statistics knowledge.
  • An interest in learning more about interactions between ocean and atmosphere.
  • Good problem-solving skills.


  • The duration is for a 8 weeks and will take place between 21 June and 1 October.
  • The project will be remote.

  • Students should submit their application by email to the CDT administrator no later than Monday 10 May 2021.

    Insertion Date: 21 April 2021


    (Back to List)

    Bayesian citizen science for STEM students at the Lake Tana Biosphere Reserve, Ethiopia

    Supervisors: Prof. Alan Blackwell of the Department of Computer Science & Technology
    and Dr Tesfa Tegegneof the Bahir Dar Institute of Technology and STEM Centre
    Project Taken

    Project Description:

    The Lake Tana Biosphere Reserve is a protected biodiversity reserve, comprising water resources and mixed-use agricultural and urban land around Lake Tana, the source of the Blue Nile and the largest lake in Ethiopia. The city of Bahir Dar is the location of the Bahir Dar Institute of Technology, which offers expertise in both AI and agricultural science.

    Bahir Dar University also hosts a STEM Centre, the first in a national network of schools established as a talent development programme that offers project-based learning for advanced high school students. The goal of the REP project is to enhance their capabilities in environmental data science, by developing educational tools that can be used for local student-led science projects in the Lake Tana Biosphere Reserve.

    Background introductory reading:
    Inventing Artificial Intelligence in Ethiopia
    A Live, Multiple-Representation Probabilistic Programming Environment for Novices
    Usability of Probabilistic Programming Languages

    Brief description of what the student will do, the skills they will gain and the outcome expected from the project: Software development in Python, resulting in a library for novel visualisation of probabilistic programming models. The student will gain expertise in data science, Bayesian statistics, user interface design.

    Please read the further particulars about the role and eligibility.

  • The ideal undergraduate could be studying CS/Maths/Engineer/Physics, or any other sciences or humanities field (e.g. biology, education or development studies), so long as the student has substantial software development experience.


  • The duration is for a 8 weeks between June and September.
  • The project will probably remote, using own laptop.

  • Students should submit their application by email to the CDT administrator no later than Monday 10 May 2021.

    Insertion Date: 21 April 2021


    (Back to List)

    Machine learning for high dynamic image merging

    Lead Supervisor: Dr. Rafal Mantiuk of the Department of Computer Science & Technology (@cam.ac.uk
    Project Taken

    Project Description:

    The goal of the project is to develop software for merging multiple images of different exposure into a high dynamic range image [1] while reducing the artifacts due to camera and scene motion. If the project is successful, it will contribute to Open Source software [2,3]. The work will require adapting the state-of-the-art machine-learning-based optical flow methods to the problem and also making them suitable for high-resolution images.

    [1] High Dynamic Range Photography
    [2] pfstools
    [3] HDRutils-pip

  • Required skills: excellent programming skills (Python, C++), fundamental knowledge of computer vision methods, familiarity with Unix environment.
  • Desirable skills and areas of expertise: PyTorch, NumPy, camera pipeline.
  • The project is planned for 8-10 weeks. The project should ideally start in June, a few days after the examinations. Other dates are also possible.
  • The work will be done either from home or from the office, depending on the restrictions due to COVID-19.

  • Students should submit their cover email with CV attached to Dr. Rafal Mantiuk

    Insertion Date: 28 April 2021


    (Back to List)

    Modelling of Net-Zero energy systems

    Lead Supervisor: Dr. Andrew Wheeler of the Whittle Lab, Department of Engineering
    Project Taken

    Project Description:

    Future Energy Systems are likely to involve a combination of closely coupled low-Carbon technologies. An example would be a combined-cycle Hydrogen-fuelled gas turbine which recovers exhaust heat for combined heat, power and cooling. In aviation, increased electrification will lead to significant heat management challenges of electrical components, and compact cooling technologies will be required.

    One of the challenges is understanding the trade-off between the additional complexity and cost of new energy systems and the likely performance benefit over the plant life-cycle. The choice of working fluid for secondary/bottoming cycles (such as steam, supercritical CO2 or an organic vapour) is also often unclear because the fluid choice affects the thermodynamic performance, the power density of the combined system and the fluid dynamic behaviour of the components.

    The project will explore how low-order models can be used to optimize complex energy systems and choice of working fluids. This will require the development of low-order models of components (such as turbomachinery, heat exchangers, pipe networks) and fluid properties databases (such as CoolProp). In recent years we have developed a number of models for the effects of working fluid properties on turbomachinery performance which we would be able to include. The aim will be to explore testcases which are relevant to both power generation and aviation.

  • The project would run over the summer of 2021 for an 8-10 week period.
  • The project is supported by MathWorks who are providing financial support for the stipend.

  • Students should submit their application by email to Dr. Andrew Wheeler

    Insertion Date: 22 April 2021


    (Back to List)

    Natural Language Processing Summer Work Projects

    Supervisors: Prof. Paula Buttery and Dr Andrew Caines from the NLIP Group, of the Department of Computer Science & Technology.
    Project Taken

    Project Description:

    We aim to select a small group of UROP students who will work together on several projects relating to educational technology, real world data and language modelling. The projects involve understanding different approaches to research questions, and the application of established natural language processing and machine learning techniques. Students will have the opportunity to develop research skills including literature search, experiment design, coding, writing and presenting.

  • We are open to applications from students of all disciplines; however, a background in computer science and/or linguistics will be an advantage.
  • 9 weeks full-time duration in the Computer Laboratory, West Cambridge; 28 June to 27 August 2021.
  • The project will be remote.

  • Please submit a 1-2 page CV to Prof. Paula Buttery and Dr Andrew Caines and explain which aspects of NLP / ML interest you; interviews (remote or in person) will be held in the Easter term.


    The project can be done remotely or in person. Insertion Date: 21 April 2021


    (Back to List)

    Modular robot kit development

    Lead Supervisor: Dr. Fumiya Iida of the Department of Engineering
    Project Taken

    Project Description:

    This project aims to develop a minimalistic robot kit that uses ARM hardware/software products, to assist students to quickly learn skills and technologies to address agri-food robotics challenges in a Project Based Learning (PBL) manner.
    In this project we proposed the development of an ARM powered open-source Agri-Food Robot Kit for research and education. The use of ARM micro-controllers will allow for quick learning and development of hardware components including connecting sensors and motors, as well as programming of computer vision, motor control, and machine learning processes.
    In this case study, we develop a minimalistic robot kit that connects simple motors and sensors (camera, and other sensor modalities) that can be used to interact with agricultural/food products.
    With the 10-week UROP project, we will develop the hardware and software components of the kit.
    The project will also be documented into an educational curriculum for undergrad students to learn in group or in individual.

  • You will be expected to programme computer vision and machine learning codes in Python, and have some basic skills of electronics hardware development.
  • The project is planned for 10 weeks and can be carried out remotely.

  • Students should submit their cover email and indicate their project preferences, with their CV attached to Dr. Fumiya Iida.
    Applications will be reviewed as they arrive, so an early submission is encouraged.

    Insertion Date: 4 May 2021


    (Back to List)

    Head-tracking for a multi-focal plane display

    Lead Supervisor: Dr. Rafal Mantiuk of the Department of Computer Science & Technology
    Project Taken

    Project Description:

    The goal of the project is to develop a method for tracking the head position with an IR camera. Such tracking is required for our prototype multi-focal plane display. The tracking method must operate in real-time with minimum latency. The method must be able to track users with and without glasses.

  • Required skills: excellent programming skills (C++), fundamental knowledge of computer vision methods, familiarity with Unix environment.
  • Desirable skills and areas of expertise: GPU programming (CUDA, OpenCL), optical flow, tracking, Kalman filter.
  • The project is planned for 8-10 weeks. The project should ideally start in July. Other dates are also possible.
  • TA part of the work will require visits to the lab so the candidate must plan to reside in Cambridge for the duration of the project. The other part of the work may need to be done from home, depending on the COVID-19 restrictions.

  • Students should submit their cover email with CV attached to Dr. Rafal Mantiuk

    Insertion Date: 28 April 2021


    (Back to List)

    Web-based interactive marching cubes

    Lead Supervisor: Prof. Graham Pullan of the Department of Engineering
    Project Taken

    Project Description:

    The Marching Cubes algorithm is a cornerstone of visualisation and imaging. From a 3-D or 2-D (marching squares!) field, the algorithm finds the triangulated surface (line in 2-D) that connects points with the same scalar value. The algorithm has found wide application in many areas of science and engineering.

    The aim of this project is to develop an efficient, interactive, web-based implementation of Marching Cubes. Interactive means that the user can change the target value and the iso-surface updates in real-time. For small datasets, this can be done on the client (web browser) via JavaScript, perhaps with acceleration from WebGL or WebGPU. For larger datasets, a different strategy will be required, likely using a WebSocket connection between the client and server.

  • This project will suit someone either with experience of software development on the client (JavaScript and WebGL) or server (likely Python) or with knowledge of related languages and an enthusiasm to learn new techniques!
  • The project is expected to run for 8-10 weeks.
  • The project can be done remotely.

  • Please submit your cover email with CV attached to Prof. Graham Pullan.

    Insertion Date: 13 May 2021


    (Back to List)

    Experimental study of insulators with Fermi surfaces

    Lead Supervisor: Dr. Suchitra Sebastian of the Cavendish Laboratory, Department of Physics
    Project Taken

    Project Description:

    Our canonical understanding of quantum condensed matter expects magnetic quantum oscillations to arise only in metals. However, our group's recent discovery of quantum oscillations in unconventional insulators (Tan et al, Science 2015; Hartstein et al, Nature Physics 2018) poses an extraordinary challenge to this 20th century convention. In this project, students will study Fermi surfaces in unconventional insulators under conditions of low temperatures and high magnetic fields. An experiment will be designed to enable the study of quantum oscillations in an insulator as a function of tilt angle of the applied magnetic field, and to thus examine the origin of the Fermi surface. This experimental work will be conducted in the Maxwell Centre at the Cavendish. In the event of covid-19 prohibiting attendance in person, students will instead be able to perform an analysis-based project from home, using data collected both in Cambridge and at high magnetic field facilities.

  • The ideal candidate will have independence, resourcefulness, good communication and teamwork skills.
  • The project will last for 10 weeks.

  • Please send an email with your CV attached to Dr. Suchitra Sebastian to register your interest or to apply.

    Insertion Date: 29 March 2021


    (Back to List)

    RF technology for robot localization in polytunnels

    Lead Supervisor: Dr. Fulvio Forni of the Department of Engineering
    Project Taken

    Project Description:

    The project will explore the use of radio sources and passive RFID tags for localization of mobile robots in polytunnels. The project will be in collaboration withDogtooth Technology Limited, whose mobile robot platform for strawberry picking is widely used around the world.

    The project will use radio signal strength and triangulation methods in conjunction with odometer information for robot localization. The goal is to guarantee wireless confinement of the robot within specific rows of the polytunnel (safety requirement) and to support navigation, with the long-term goal of enabling vision-based technology.

  • Good programming skills are required (Arm Mbed-OS compatible microcontroller, Python).
  • The student will have the opportunity to work directly on the Dogtooth's robot platform.
  • Previous experience with RF technology is a plus.
  • This is a 10-week project. The project must be carried out in person during the summer.

  • Students should submit their cover email with CV attached to Dr. Fulvio Forni

    Insertion Date: 7 May 2021


    (Back to List)

    Advanced control algorithms for robotics

    Lead Supervisor: Dr. Fulvio Forni of the Department of Engineering
    Project Taken

    Project Description:

    The project will develop advanced feedback control algorithms for robotics. The goal of the project is to build a library of models and algorithms to support student research and education. The project will be in collaboration with Mathworks.

    The project will focus on control of robotic manipulators, within the context of full actuation and under-actuation. It will also focus on interaction control, within the context of contact modelling and haptic interfaces. Locomotion will also be considered (walking/jumping robots).

  • Simscape Multibody will be used for modelling.
  • Control algorithms will be implemented in Simulink.
  • Code generation for VEX-robotics platform could also be considered.
  • This is a 10-week project. The project can be done in person or remotely during summer (in person is recommended).

  • Students should submit their cover email with CV attached to Dr. Fulvio Forni

    Insertion Date: 7 May 2021


    (Back to List)

    Atmospheric Evolution of Exoplanets

    Lead Supervisor: Dr. Paul B. Rimmer of the Department of Earth Science
    Project Taken

    Project Description:

    How do atmospheres of rocky planets evolve? This project will explore the evolution of rocky planet atmospheres, from the initial nebular gas to impact-generated atmospheres, volcanic atmospheres and biogenic atmospheres.

    The project will focus on generalizing an atmospheric evolution model that is coupled to an atmospheric chemistry model. The student will learn how to run the models, how to adjust input and analyze output, and then apply the model to example systems of the student’s interest: for example, early Earth, Venus, Mars, an impact-eroded atmosphere, a hot volcanic rocky exoplanet (GJ 1132 b).

  • The atmospheric evolution code is written in Python3.
  • The atmospheric chemistry code is written in Python2.7, C++ and FORTRAN 77.
  • This is primarily an atmospheric chemistry/geochemistry focused project.
  • This is an 8-week project. The project can be done in person or remotely during summer.
  • The Simons Foundation will be providing financial support for the stipend.

  • Students should submit their cover email with CV attached to Dr. Paul B. Rimmer
    Applications will be reviewed as they arrive, so an early submission is encouraged.

    Insertion Date: 11 May 2021


    (Back to List)

    The UV Environment of Mars and Exo-Mars

    Lead Supervisor: Dr. Paul B. Rimmer of the Department of Earth Science
    Project Taken

    Project Description:

    There are several challenges to modelling the atmospheric chemistry of Mars and other low-surface-pressure rocky planets. Scattering by cloud aerosols and temperature-dependent ultraviolet chemistry become significant, and must be accounted for. The result of this work will be applied to future Mars research and exoplanet research carried out in Cambridge, Bern, and Prague.

    The project involves updating a chemical kinetics code for Python3, and then working with a new radiative transfer code that can account. for cloud aerosols. The bulk of the project will involve developing a computational solution to incorporate temperate changes in ultraviolet absorption cross-sections.

  • The atmospheric chemistry code is written in Python2.7, C++ and FORTRAN 77.
  • This is primarily a computer programming project, but also involves some planetary science (primarily for Mars and high temperature rocky exoplanets) and atmospheric chemistry.
  • This is an 8-week project. The project can be done in person or remotely during summer.
  • The Simons Foundation will be providing financial support for the stipend.

  • Students should submit their cover email with CV attached to Dr. Paul B. Rimmer
    Applications will be reviewed as they arrive, so an early submission is encouraged.

    Insertion Date: 11 May 2021


    (Back to List)

    Machine learning for soft robot simulation

    Lead Supervisor: Dr. Fumiya Iida of the Department of Engineering
    Project Taken

    Project Description:

    This project develops a set of simulation models for the purpose of effective and efficient student research and education for robotics and machine learning. The models consist of mechanics simulation of complex soft robots, and machine learning toolboxes applied to control them. By building and experimenting with these systems, the students should be able to learn the fundamentals of robot design, control, and simulation, both theoretically and practically. The student working on this project

  • You will need to work with basic matlab scripts, SimuLink and other toolboxes, Live Scripts and possibly CAD designs.
  • A good technical writing skill for instructions and reporting is also necessary.
  • The project is planned for 10 weeks and can be carried out remotely.

  • Students should submit their cover email and indicate their project preferences, with their CV attached to Dr. Fumiya Iida.
    Applications will be reviewed as they arrive, so an early submission is encouraged.

    Insertion Date: 4 May 2021


    (Back to List)

    Spatial mechanotrancriptomics of the mouse embryo

    Supervisors: Dr. Bianca Dumitrascu, Group Leader, Department of Computer Science & Technology
    Dr Adrien Hallou, Herchel Smith Research Fellow, Wellcome Trust/CRUK Gurdon Institute & Cavendish Laboratory
    Project Taken

    Project Description:

    What is the nature of the mechanisms leading to the formation of a highly complex and spatially organised organism from a single cell is the central question of developmental biology. To better understand the complex interplay between genomic information and the morphogenetic movements which sculpt the developing embryo in space and time, we propose to work on a new and unique approach combining mechanical force inference with the image-based single-cell transcriptomics method, seqFISH. Using this approach, we will simultaneously and precisely infer for each single cell of a 8-12 somite stage mouse embryo its mechanical state and the gene expression profile for 387 selected target genes. This will allow us to unravel correlations between gene expression profile and mechanical forces at the cellular, tissue and organism scale and then, using machine learning approaches, to compute pseudotime trajectories in order to quantify in space and time the role of mechanical forces on gene expression patterns and cell fate decisions.

  • Duration 8 weeks.
  • The project can be conducted entirely remotely if needed.
  • Essential skills: The student should be comfortable with scientific computing and programming in Python.
  • Other skills: Knowledge of R, bioinformatics tools and machine learning would be really helpful, as well as a keen interest for physical and quantitative approaches of biological phenomena.
  • The student subsistence costs will be funded by the the Cambridge Centre for Physical Biology.

  • Deadline for applications have been extended until 9 June 2021.

    For information regarding the application process please check the CPB website

    Insertion Date: 3 June 2021


    (Back to List)

    Robustness of multi-tissue elongation in a chick embryo

    Supervisors: Dr. Fengzhu Xiong of the Wellcome Trust / CRUK Gurdon Institute
    Project Taken

    Project Description:

    A developing embryo is like a jigsaw puzzle where distinct pieces (tissues) must fit closely to ensure an overall correct structure. What mechanisms coordinate the sizes and shapes of different tissues are unclear. Our earlier work on the early chicken embryo shows that tissues in the elongating body axis shape each other through mechanical forces. Recently, we found that these forces appear to respond to changing tissue shapes as a feedback to minimize deviation from the target shape, raising a possible mechanism of robustness. In this project the student will have the opportunity to identify the cellular basis of this responsive mechanism with a combination of data analysis, modeling and experiment.

    Key aims and tasks:
    (1) Quantifying tissue deformation and cell dynamics of shape perturbed embryos;
    (2) Constructing a model to predict force changes with the quantitative measurements;
    (3) Testing the predictions in live avian embryos with mechanical and surgical tools.

  • Duration 8 weeks.
  • Lab-based project but some components can be performed remotely if need be.
  • Skills required: Basic knowledge of cell and developmental biology, mechanics. Math and programming skills are advantageous but not absolutely required.
  • The student subsistence costs will be funded by the the Cambridge Centre for Physical Biology.

  • Deadline for applications have been extended until 9 June 2021.

    For information regarding the application process please check the CPB website

    Insertion Date: 1 June 2021


    (Back to List)

    RNA structure in health and disease - investigation of lncRNA structure function relationship in imprinting control regions

    Supervisors: Dr Russell S. Hamilton of the Department of Genetics
    Project Taken

    Project Description:

    Long non-coding RNAs (lncRNAs) are fundamental to the function of clusters of imprinted genes, however little is known about their structural features, likely to be key to understanding their functional roles and links to human disease. Here we propose to structurally characterising 5 key imprinted lncRNAs alongside an extensive assessment of their key structural features. These lncRNAs have been shown to have key roles in placenta and embryo development, as well as an emerging role in brain development and function.

    Key aims and tasks:
    (1) Generate accurate 3D models of imprinted lncRNAs
    (2) Perform sequence (1D), 2D and 3D comparison across the lncRNAs
    (3) Assess 1D/2D/3D conservation between human and mouse
    (4) Present results in a publicly available web resource

    For more information see the full proposal.

  • Duration 8 weeks.
  • The project is purely computational so can be conducted remotely. If restrictions allow the project can be conducted in within the Department.
  • Skills required: Essential skills: basic competency in R and using bioinformatics tools, although training will also be provided.
  • The student subsistence costs will be funded by the the Cambridge Centre for Physical Biology.

  • Deadline for applications have been extended until 9 June 2021.

    For information regarding the application process please check the CPB website

    Insertion Date: 1 June 2021


    (Back to List)

    Live Modelling the Vertebrate Presomitic Mesoderm

    Supervisors: Dr. Ben Steventon of the Department of Genetics
    Co-supervisors: (Wet Lab) Mr Tim Fulton, Department of Genetics and (Mathematical Modelling) Dr. Berta Verd, Department of Zoology, University of Oxford.
    Project Taken

    Project Description:

    Pattern formation in tissues also undergoing morphogenesis requires a cell to dynamically update its identity with respect to its changing position.

    The aims of this project are to:
    (1) Investigate how morphogenetic perturbation of embryos, through inhibition of cell movements, results in changes to the patterns produced both experimentally and in silico.
    (2) Investigate the translatability of a conserved gene regulatory network onto tracks from related species (Cichlids)

    For more information see the full proposal.

  • Duration 8 weeks.
  • If permitted, we intend for the student to get hands on laboratory experience. If this is not possible, the entire project can be adapted for a working from home style project.
  • Some knowledge of coding using Python would be a significant advantage for the student. Any lab techniques will be taught and no previous lab experience is required.
  • The student subsistence costs will be funded by the the Cambridge Centre for Physical Biology.

  • Deadline for applications have been extended until 9 June 2021.

    For information regarding the application process please check the CPB website

    Insertion Date: 1 June 2021


    (Back to List)

    Biological Data Engineering for Geometric Deep Learning

    Supervisors: Prof Pietro Lio of the Department of Computer Science & Technology
    Co-supervisor: Arian Jamasb of the Department of Computer Science & Technology
    Project Taken

    Project Description: Graphein is a python library for facilitating geometric deep learning research in Biology. The library provides a suite of functionality for creating and representing data for deep learning projects from a variety of biological sources such as protein structures, interaction networks and chromatin structure data. The project is to develop further featurisation schemes, allowing users to build richer representations of biological data for deep learning tasks and to increase the number of data modalities the library can support.

    We currently have a large number of users and are working to position this library as the gold-standard for processing biological graphs in the context of deep learning. We are also interested in creating rich graph objects, such as hierarchical, hyper and multigraphs and are happy for a motivated student to carve out their own project based on their interests.

    This is an excellent project for a student looking to increase their exposure to deep learning and computational biology.

    Key aims of this project are to increase the number of featurisation schemes and data modalities supported by our data processing library. www.github.com/a-r-j/graphein

  • 6-8 weeks depending on the student's availability.
  • The project can be conducted entirely remotely.
  • Students should be comfortable with python, scientific computing and git. The student can expect to gain knowledge of deep learning (specifically geometric deep learning - a frontier of the field with lots of early successes in biology & life sciences) and their associated frameworks. The student will also learn good software development practices.
  • The project is purely computational so can be conducted remotely. If restrictions allow the project can be conducted in within the Department.
  • The student subsistence costs will be funded by the the Cambridge Centre for Physical Biology.

  • Deadline for applications have been extended until 9 June 2021.

    For information regarding the application process please check the CPB website

    Insertion Date: 1 June 2021


    (Back to List)

    The role of CTCF and chromosome topology in the regulation of gene expression in imprinted domains

    Supervisors: Dr Carol Edwards of the Department of Genetics
    Project Taken

    Project Description:

    The DNA of cells is exquisitely folded within the nucleus. Topologically associated domains (TADs) are self-interacting regions on a chromosome that are demarcated by CTCF binding and thought to constrain gene regulation to within the TAD.

    This project aims to study the role of topology in the control of gene expression using genomic imprinting as a model. We will explore the impact of deleting CTCF sites at the edge an imprinting domain on gene expression, imprinting and topology.

    Key aims and tasks:
    (1) Extract RNA from tissues dissected from mice heterozygous for CTCF binding site deletion.
    (2) Perform quantitative PCR to assess expression levels in maternal and paternal heterozygotes and their wildtype littermates.
    (3) Assess imprinting using allele specific pyrosequencing across known polymorphisms then compare these results with those from other models.
    (4) Use circular chromosome conformation capture sequencing (4Cseq) to assess changes to topology in the region.

  • This project will be co-supervised by Prof. Anne Ferguson-Smith of the Department of Genetics.
  • Duration 8 weeks.
  • In person: The project is lab based but some analysis can be done remotely.
  • Skills required: Basic knowledge of molecular biology and genetics. Some lab experience is desirable but full training will be given.

  • Please read the further particulars about the role and eligibility.

    The student subsistence costs will be funded by the the Cambridge Centre for Physical Biology.

    Deadline for applications have been extended until 9 June 2021.

    For information regarding the application process please check the CPB website

    Insertion Date: 26 May 2021


    (Back to List)

    Variability in diameter of axonal endoplasmic reticulum tubules

    Supervisors: Dr Cahir O'Kane of the Department of Genetics
    Project Taken

    Project Description:

    Endoplasmic reticulum (ER) forms a continuous intracellular tubular nanostructure along the length of axons. Axonal ER has an unusually narrow diameter, which constrains diffusion along its length; larger tubules, caused by loss of genes whose loss causes axon degeneration disease, allow faster diffusion. We will explore how uniformly this diffusion constraint is distributed along tubules, by (1) measuring how tubule diameter varies along their length, and (2) developing tracking of single molecules inside tubules.

    Key aims and tasks:
    (1) Use existing serial EM sections to manually measure individual tubule diameters along successive sections.
    (2) Assess parameters, e.g. variance, interquartile range, to describe variability along length of individual tubules, and can be used to statistically compare variation across tubules, axons and genotypes.
    (3) Test object segmentation and automated measurements, for consistency with manual measurements.
    (4) Laboratory access permitting, use confocal microscopy to test efficiency of in vivo labeling and diffusion of lumenal markers

    Please read the further particulars about the role.

    The student subsistence costs will be funded by the the Cambridge Centre for Physical Biology.

  • Duration 8 weeks.
  • The project has a major component that can be conducted remotely, and a component that is lab-based, if distancing/occupancy constraints allow lab access over the summer. If lab access is not practicable, the entire project can be remote.
  • Essential skills: numeracy, basic statistical knowledge
  • Desirable experience, but training will be given: image analysis; knowledge of subcellular organisation; fluorescence/confocal microsopy; statistical software; basic genetics

  • Deadline for applications have been extended until 9 June 2021.

    For information regarding the application process please check the CPB website

    Insertion Date: 24 May 2021


    (Back to List)

    Tracking individual bacterial cells and estimating their age in a growing biofilm

    Supervisors: Dr Somenath Bakshi of the Department of Engineering
    Project Taken

    Project Description:

    Recent experiments have shown that even symmetrically dividing bacteria can undergo aging. There is also preliminary evidence that suggests aging can have important implications in affecting the ability of bacteria to tolerate antibiotics. In this project, we would like to investigate how bacteria age as they form biofilms, which are colonies of bacteria that are major contributors to recalcitrant infections.

    Specific aims:
    (1) We will start with using computational models of biofilm (developed in our lab and elsewhere) to analyse replicative aging of cells in model biofilms. Replicative aging refers to the number of daughters produced from a cell.
    (2) We will then use the biofilm imaging platform developed in the lab to image biofilm development from individual cells with single-cell resolution.
    (3) Next, we will develop an image analysis pipeline to track individual cells in the images of the growing biofilm and estimate their age from the number of divisions undergone. The results will be compared with results from the model in step 1, and the model will be updated accordingly.

    The student subsistence costs will be funded by the the Cambridge Centre for Physical Biology.

  • Duration 8 weeks.
  • The project is a good mix of computational work and experiments (70:30). The modelling work can be done remotely. We are quite confident about lab access, but in the unlikely event of another lockdown, we can provide the student with data for image analysis to directly work on step 3, which can be done remotely.
  • Useful knowledge, skills and attributes: Image analysis, programming, statistics, basic microscopy and microbiology
  • Recommended references: Microbial ageing and longevity and Biased partitioning of multi-drug efflux pumps help older cells to tolerate antibiotics.

  • Deadline for applications have been extended until 9 June 2021.

    For information regarding the application process please check the CPB website

    Insertion Date: 24 May 2021


    (Back to List)

    An assay to resolve dynamic fitness costs of gene-expression at single cell level

    Supervisors: Dr Somenath Bakshi of the Department of Engineering
    Project Taken

    Project Description:

    Fitness costs of gene-expression are of fundamental interest since these dictate the ecology and evolution of the species. The 'cost' of expressing a gene is dictated by the required amount of cellular resources, which are otherwise invested in cellular growth and maintenance, often termed as metabolic burden. We plan to develop a new assay to quantify metabolic burden of dynamic gene-expression in individual cells, to estimate the corresponding fitness costs and compare with bulk measurements.

    In this project, we plan to run pilot experiments with an inducible gene to switch between 'on' and 'off' states and develop an analysis pipeline to compute the fitness cost of the 'on' state. In addition to this, as an orthogonal control, we plan to build a platform to image and count colonies from cells from a bulk culture to estimate the fitness differences from the changes in their relative abundance over time.

    Specific aims:
    (1) Compute the fitness costs of expression of a gene by turning it 'on' and 'off' using an inducer and estimating relative growth-rates of individual cells in the microfluidic device.
    (2) Construct a plate imaging platform to count colonies from a mixed culture containing cells from strains with and without this gene (constantly 'on' or 'off').
    (3) Compute the relative fitness of these two strains using the relative colony counts and compare with results from aim 1.

    Please read the further particulars about the role.

    The student subsistence costs will be funded by the the Cambridge Centre for Physical Biology.

  • Duration 8 weeks.
  • The project is primarily lab based but will include tasks that can be done remotely.
  • Useful knowledge, skills and attributes: microbiology, microscopy, instrumentation, image analysis, and statistics.

  • Deadline for applications have been extended until 9 June 2021.

    For information regarding the application process please check the CPB website

    Insertion Date: 20 May 2021


    (Back to List)

    The mechanics of biofilm morphogenesis

    Supervisors: Dr Nuno Miguel Oliveira of the Department of Applied Mathematics and Theoretical Physics and Veterinary Medicine
    Project Taken

    Project Description:

    Microbiology has traditionally focused on how bacteria respond to their chemical environment and, therefore, our understanding of their mechanobiology is very limited. In particular, bacteria typically live as surface-associated communities (biofilms) where they experience a range of forces, but these have remained elusive. This project aims to quantify these forces by studying biofilms of bioluminescent bacteria, whose light emission is triggered by mechanical stimuli.

    The summer student will analyse experimental data where bacteria were grown on different stiffness substrates, which generate light patterns, and will develop a mathematical model that can help to explain the observed patterns.

    Please read the further particulars about the role.

    The student subsistence costs will be funded by the the Cambridge Centre for Physical Biology.

  • Duration 8 weeks.
  • Project can be done remotely.
  • Experimental work might be possible depending on pandemic-related restrictions at the time.
  • The ideal candidate should be familiar with quantitative and image analysis, programming, fluid mechanics and pattern formation.

  • Deadline for applications have been extended until 9 June 2021.

    For information regarding the application process please check the CPB website

    Insertion Date: 20 May 2021


    (Back to List)

    Reduced power measurements of Drosophila larvae for 2-photon optogenetic stimulation of neurons

    Supervisors: Dr Miranda Robbins of the Department of Zoology
    Project Taken

    Project Description:

    This project aims to identify the optimal immobilisation method for 2-photon imaging in live samples using a novel biomaterial. Tasks include optimising an embedding method for live Drosophila larvae. Designing a 2-photon stimulation protocol to photoactivate fluorescent proteins in individual neurons. Comparing the optical properties of several embedding materials. The student will then use existing methods or develop their own image analysis pipeline to quantify the resulting data.

  • This project will be co-supervised by Dr Marta Zlatic of the Department of Zoology.
  • Duration 6 weeks (5 weeks wet lab, 1 week computational analysis).
  • The project requires wet lab techniques. Depending on government regulations at the time, it is possible to perform the project remotely through learning or developing image analysis methods.
  • No essential previous skills are required, however it would be desirable for students to have a combined interest in optical microscopy, biotechnology, and neuroscience.

  • Please read the further particulars about the role and eligibility.

    The student subsistence costs will be funded by the the Cambridge Centre for Physical Biology.

    Deadline for applications have been extended until 9 June 2021.

    For information regarding the application process please check the CPB website

    Insertion Date: 25 May 2021


    (Back to List)

    Visualisation tool for interaction between machine learning and breast cancer data analysis

    Lead Supervisor: Prof. Pietro Lio, Prof. Mateja Jamnik and Dr. Nikola Simidjievski of the Department of Computer Science & Technology
    Project Taken

    Project Description:

    The aim of this project is to build on and extend the VIIDA prototype: Visual Interface for Integrative Data Analysis. The prototype follows the latest trends in data visualisation and human-computer interaction and implements various data analysis workflows. It needs to be built upon to extend the number of machine learning models available for analysis, the interaction capabilities for data analysis, and the integration of different data types, in particular medical images.

  • The ideal candidate for this project will have a great JavaScript (and various open-source libraries eg. React, Plotly etc.) and Python programming skills.
  • Familiarity/interest in biostatistics and machine learning is a plus.
  • The project is planned for 8-10 weeks. The project can be done in person or remotely but a few, regular, in person meetings will be required.
  • The work will be done either from home or from the office, depending on the restrictions due to COVID-19.

  • Please submit your CV to Prof. Pietro Lio or Prof. Mateja Jamnik

    Insertion Date: 11 May 2021


    (Back to List)

    Kinetics of CNT Reactors

    Supervisor: Dr. Adam Boies, Department of Engineering
    Project Taken

    Project Description:

    The production of solid carbon from reactions may serve as a means to limit CO2 emissions, making carbon more valuable than a waste product. This project seeks to investigate the fundamental kinetics, and thus limits in to solid carbon (carbon nanotube, CNT) production from a floating catalyst chemical vapor deposition (FC-CVD) reactor. This work provides critical kinetics measurements to understand the limits to the growth processes for optimization of reactor design and catalyst production by investigating the measurement of CNT growth rates from catalysts. These efforts will combine new capabilities at Cambridge for extraction of reactor species from the CVD furnace for real-time measurement of CNT mass. Newly precured equipment for both gas and particle measurement will enable high resolution measurements of gas, catalyst and CNTs that were not previously possible.


  • The work will be experimental in nature and it is expected that the UROP student will have a good knowledge of thermofluids.
  • The project will run for 10 weeks and is flexible in terms of start and end date.
  • Please apply to the lead Supervisor, Dr. Adam Boies.

    Insertion Date: 16 June 2021

  • (Back to List)

    Anomalous adhesion in liquid crystalline elastomers

    Lead Supervisor: Professor Eugene M. Terentjev of the Department of Physics
    Project Taken

    Project Description:

    This project is to run alongside the group active research into liquid crystalline elastomers (LCEs), which are a unique class of polymer materials with mechanical properties unlike any other solid. They that show spontaneous and reversible shape changes of up to 500%, produced on heating, or light exposure. They also have the soft elasticity when some shear deformations occur without any elastic resistance, which in turn leads to their remarkable properties in damping and dissipation of mechanical energy. Our group have been the world leaders in the field for over 20 years; chemists produce a range of conceptually new materials to study, and physicists investigate their properties and design their applications.

    One such unusual LCE property is the dynamic adhesion, when the elastomer is sticky in the liquid-crystal phase, but is not sticky at all in the low-temperature glassy—and in the high temperature isotropic phases. This project will focus on this effect: characterising the adhesion strength depending on various material and environmental parameters. This is an exciting off-shoot of our regular projects, which we need the help with this summer - hopefully it will lead to a completed paper.

  • This project will be co-supervised by Dr Mohand Saed of the Department of Physics.
  • This project is suitable for a candidate with interest and a good set of laboratory skills in physics/materials or (micro)mechanical engineering.
  • The amount of work we have under this project is almost infinite, so the summer project could easily be continued into a Part III research project or further (if all goes well).

  • Please apply to the lead Supervisor, Professor Eugene M. Terentjev.

    Insertion Date: 25 May 2021


    (Back to List)

    Visualising Data Bias

    Supervisor: Dr. Marcus Tomalin, Department of Engineering
    Project Taken

    Project Description:

    This project will focus on the task of creating an interface for NLP text corpora that enables different kinds of bias to be revealed visually and interactively prior to the data being used to train systems. The interface would constitute a 'dynamic data statement' that would extend the static 'data statement' framework introduced in Bender & Friedman 2018 so that it includes elements of data visualisation. This research will form part of a joural article about visualising data bias and the student who undertakes this research will be a named author in that article.


  • A secure knowledge of Python is essential.
  • An interest in NLP techniques and familiarity with NLP tools (e.g., NLTK) is highly desirable
  • The research can be undertaken in Cambridge or remotely.
  • Please apply to the lead Supervisor, Dr. Marcus Tomalin.

    Insertion Date: 16 June 2021

  • (Back to List)

    Probing the Cosmic Dawn with a long wavelength radiometer

    Lead Supervisor: Dr. de Lera Acedo of the Department of Physics
    Project Taken

    Project Description:

    In this project the student will work withing the REACH team, exploring the formation of the very first stars and how they shaped the Universe we know today by studying radio signals emitted by the raw material that formed those stars (Neutral Hydrogen).

    Overview of tasks:

  • Development of physics motivated models of the foreground signals that shadow the cosmological signal of interest
  • Analysis of simulated or real telescope data using REACH's Data analysis pipeline
  • Interpretation of the results in the light of state of the art theoretical models

  • The ideal candidate should be enthusiastic about cosmology and have good programming skills.
  • Location: Department of Physics in West Cambridge.
  • Duration: 10 weeks.
  • This project could lead into a final year project on similar topics.

  • Please send an email with your CV attached to Dr. de Lera Acedo to register your interest or to apply.

    Insertion Date: 9 March 2021


    (Back to List)

    Learning to predict Carbon measurements from NASA/ESA remote sensing data

    Lead Supervisor: Dr Cengiz Oztireli of the Department of Computer Science & Technology
    Project Taken

    Project Description:

    We have an archive of weekly remote sensing data for the world, spanning decades, mainly from ESA and NASA. The work would involve combining these data with ground readings for variables such as Soil Organic Carbon to train models that can output Carbon measurements using only remote sensing data. We can also look at incorporating data from the latest generation of hyperspectral sensors to see how this improves model accuracy.

  • The project is planned for 9 weeks.
  • The project can be done in person or remotely.

  • Please submit your cover email with CV attached to Dr Cengiz Oztireli.

    Insertion Date: 14 May 2021


    (Back to List)

    Security Aspects of Brain Computing Interfaces (BCIs)

    Lead Supervisor: Dr. Anil Madhavapeddy of the Department of Computer Science & Technology
    Co-supervisors: Zahra Tarkhani and Lorena Qendro
    Project Taken

    Project Description:

    Brain Computing Interface (BCI) technologies, both invasive and non-invasive, are increasingly used in a wide range of applications, from health-care to smart communication and control. Most BCI applications are safety-critical or privacy-sensitive. However, the infinite potentials of BCI and its ever-growing market size have been distracted the BCI community from significant security and privacy threats. In this research, we investigate the security and privacy threats of various BCI devices and applications, from machine learning adversarial threats to untrusted systems and malicious applications.

    In this project, you explore the impact of current security threats on BCI stacks, including applications, frameworks, libraries, and systems abstractions. You will also investigate the possibility of new attack vectors and build tools to make the security analysis easier and more fun/automatic.

  • Essential knowledge: Development skills with C/C++ and scripting languages (e.g., Python).
  • Experience with embedded devices, OS and sandboxes, reverse engineering, and threat analysis is preferred.
  • The project is planned for 10 weeks.
  • The project can be done remotely.

  • To apply, please submit your CV to Zahra Tarkhani.

    Insertion Date: 23 June 2021


    (Back to List)

    Adversarial attacks on BCIs (Brain Computing Interfaces)

    Lead Supervisor: Dr. Anil Madhavapeddy of the Department of Computer Science & Technology
    Co-supervisors: Zahra Tarkhani and Lorena Qendro
    Project Taken

    Project Description:

    Brain Computing Interface (BCI) technologies, both invasive and non-invasive, are increasingly used in a wide range of applications, from health-care to smart communication and control. Most BCI applications are safety-critical or privacy-sensitive. However, the infinite potentials of BCI and its ever-growing market size have been distracted the BCI community from significant security and privacy threats. In this research, we investigate the security and privacy threats of various BCI devices and applications, from machine learning adversarial threats to untrusted systems and malicious applications. In this project, you explore various methods to detect and analyze security threats on BCI ML models, including attacks based on perturbed inputs, inference, and model patterns.

  • Essential knowledge: Development skills (e.g., C, C++, Python) and experience with at least one ML/Deep Learning framework such as PyTorch or TensorFlow.
  • Previous work on embedded devices and adversarial attacks is preferred.
  • The project is planned for 10 weeks.
  • The project can be done remotely.

  • To apply, please submit your CV to Zahra Tarkhani.

    Insertion Date: 23 June 2021


    (Back to List)

    Differentiable/neural rendering of unstructured point clouds

    Lead Supervisor: Dr Cengiz Oztireli of the Department of Computer Science & Technology
    Project Taken

    Project Description:

    You will be implementing differentiable/neural rendering of point clouds and solve inverse problems with the renderer.
    1. Understanding what differentiable/neural rendering is and how it works.
    2. Downloading and installing a deep learning framework.
    3. Forking the codebase of an existing neural renderer.
    4. Running experiments, extending the existing renderer with shading models.
    5. Implementing an optimization-based algorithm for an image-to-geometry inverse problem.
    6. Testing the implementation for novel scenes.

  • You will need excellent programming skills in Python, familiarity with Computer Vision and Machine Learning.
  • The project is planned for 6 weeks.
  • The project can be done in person or remotely.

  • Please submit your cover email with CV attached to Dr Cengiz Oztireli.

    Insertion Date: 19 May 2021


    (Back to List)

    BELOW THIS LINE IS NOT VISIBLE ON THE WEB SITE So copy and paste the text and hyperlinks between the lines and use as template DELETE WORDS BETWEEN asterisks.** __________________________________________________________________________ PROJECT TAKEN IMAGES Project Taken

    Project Available

    ++++++++++ DUMMY PROJECT DESCRIPTION +++++++++++++

    NAME OF THE PROJECT

    Contact:Lead Supervisor: Dr.Zuess Secret Lab
    Project Available

    Project Description:

    Setting the world to rights.


    Please apply to Contact/Lead Supervisor.

    This project will be co-supervised by Professor Bob Dillon Cavendish Lab

    Working 10 weeks over the summer

    Insertion Date: 1 February 2016


    (Back to List) _______________________________________________________________________________________________