Join

RTSRoutes in

Getting into AI for science at Cambridge

The route in is specific: a module, a project, a group and a deadline. This page names all four, for undergraduates choosing options and for anyone applying for a masters or a PhD.

Checked Sun 25 Oct 2026 against departmental and admissions pages. Confirm on the official page before you apply.

§1Deadlines

The deadline that matters is in December, not May.

Funding is decided months before course places. Applying in May usually means applying without funding.

  1. closed

    Gates Cambridge, United States round

    US citizens resident in the US. Everyone else uses the dates below.

  2. University Postgraduate Funding Competition, first round Funding

    Most PhDs, the MPhil MLMI, the BBSRC DTP and NanoDTC's first review. This round reaches the Cambridge Trust, Harding and departmental studentships.

  3. University Postgraduate Funding Competition, second round Funding

    MASt Mathematics, MPhil Scientific Computing, MPhil Computational Biology and the Sensor CDT.

  4. Most course deadlines

    Departmental PhDs and several MPhils. Applying this late usually means applying without funding.

Harding scholarships are open to any discipline and any nationality, through the same December and January funding rounds. Every deadline is also in the community calendar.

§2Routes

Courses, programmes and groups

Modules

Four modules are written for AI for science inside their own disciplines, marked Flagship. Start with the one for your subject.

Engineering

  • 3F8

    Inference

    Core machine learning: regression, classification, PCA, mixtures and EM, hidden Markov models.

    Part IIA
  • 4F13

    Probabilistic Machine Learning

    Gaussian processes, variational inference and expectation propagation, assessed by coursework.

    Part IIB
  • 4C11

    Data-driven and learning based methods in mechanics and materials Flagship

    Physics-informed networks, neural operators and generative models, with all coursework in PyTorch.

    Part IIB

Chemistry

  • Programming option in Python

    Self-taught with drop-in support. The first coding credit for chemists.

    Part II
  • L7

    Machine learning in atomistic simulation Flagship

    Descriptors, machine-learned potentials and active learning, taught by the people who built MACE. Take B7 first.

    Part III

Materials

  • M15

    Materials Informatics Flagship

    Databases, DFT-based models, regression and machine learning, and structure prediction.

    Part III

Physics

  • Computational Physics and its project

    Part III has no taught machine-learning option, so the research project is the route in.

    Part II

Maths

  • Mathematics of Machine Learning

    Stochastic gradient descent, random forests, boosting and neural networks. There is no Part III version.

    Part II

  • Statistical Learning in Practice

    Applied machine learning in R and Python, taught in Lent.

    Part III

Computer Science

  • Machine Learning and Bayesian Inference

    The core probabilistic machine-learning course, taught in Lent.

    Part II
  • L48

    Machine Learning and the Physical World Flagship

    Emulation, experimental design and multifidelity models, with a 70% group project. The most transferable course here.

    Part III / MPhil ACS

Biology

  • Systems Biology

    Python and R, omics and network modelling, with a 12-week project. Open to four triposes.

    Part III

L65 Geometric Deep Learning is not running in 2026-27. Computer Science listings are the published 2025-26 syllabus.

Masters

Taught masters that are real routes into AI for science, in order of deadline. Course deadlines are later than funding deadlines.

  • Engineering11 months

    MPhil in Machine Learning and Machine Intelligence

    The core ML masters, with five tracks including a new Biological Learning track.

    Apply by

  • DAMTP and DPMMS9 months

    MASt in Mathematics (Part III)

    Access to every Part III statistics, data-science and applied-mathematics course.

    Apply by

  • DAMTP, with C2D3 Computational Biology10 months

    MPhil in Computational Biology

    Bioinformatics, scientific programming, genomics and machine learning, with a three-month project.

    Apply by

  • Computer Science and Technology9 months

    MPhil in Advanced Computer Science

    Five taught modules including L48, and a two-term research project.

    Apply by

  • Physics, with DAMTP and the Institute of Astronomy10 months

    MPhil in Data Intensive Science

    Statistical data analysis, machine learning and research computing. Built for physicists.

    Apply by

  • Physics, Centre for Scientific Computing12 months

    MPhil in Scientific Computing

    Half taught and half research, with continuum, atomistic and HPC streams.

    Apply by

Doctoral routes

Departmental PhDs and the doctoral training centres still recruiting. A training centre adds a cohort and a first year of courses.

  • Departmental PhD

    Engineering, Physics, Chemistry, Computer Science, Materials, DAMTP, DPMMS

    Usually the best route: you choose the supervisor and the project.

    Deadline13 May 2027; funding by 8 Dec 2026

  • FIBE3: Future Infrastructure and Built Environment

    Cambridge Engineering

    MRes year then a three-year PhD. Replaces FIBE2, which has closed.

    Deadline14 Apr 2027

  • Sensor CDT

    Chemical Engineering and Biotechnology

    The MRes includes a machine-learning module, with two supervisors from different disciplines.

    Deadline6 Jan 2027 international; 13 May 2027

  • Materials 4.0

    Henry Royce Institute and Manchester, with Cambridge

    Digitalising materials research, with a machine-learning training week.

    Deadline13 May 2027

  • UKRI AI CDT in Decision Making for Complex Systems

    Manchester-led, Cambridge route via Computer Science

    Uncertainty, human-in-the-loop decisions and interpretability.

    DeadlineApply to Manchester first

  • NanoDTC

    Cambridge

    A 3.5-year PhD with six months of hands-on training. Computational projects are possible.

    DeadlineReviews 8 Dec 2026 and 2 Jan 2027

  • Cambridge Biosciences BBSRC DTP

    With Babraham, Sanger, EMBL-EBI, NIAB and Kew

    Rotations or iCASE, with training in statistics, programming and data analysis.

    Deadline8 Dec 2026

Closed or replaced. AI4ER is no longer recruiting further intakes. FIBE2 is superseded by FIBE3, and C-CLEAR by the CREATES Doctoral Landscape Award.

New this year. The Cavendish New Frontiers Programme took its first cohort in October 2026. It trains Physics PhD students in AI-native scientific workflows through standard PhD admissions.

Research groups

Showing 31 of 31 groups

  • Machine Learning Group and the Computational and Biological Learning Lab

    Ghahramani, Rasmussen, Turner, Hernández-Lobato, Weller, Ge

    Probabilistic machine learning, machine learning for weather and the Earth system, generative models for molecules.

    EngineeringEngineeringCross-cutting

  • Gábor Csányi

    Machine-learned interatomic potentials and the MACE-MP-0 foundation model.

    Engineering; Chemistry Theory groupChemistryMaterials

  • Mark Girolami

    Data-centric engineering, statistical finite elements and digital twins.

    Engineering (Civil)Engineering

  • Matthew Juniper

    Bayesian inference, adjoint methods and data assimilation into physics-based models.

    Engineering (Energy, Fluids and Turbomachinery)EngineeringPhysics

  • Whittle Laboratory

    GPU-accelerated turbomachinery simulation, net-zero aviation and rapid testing with AI.

    EngineeringEngineering

  • Centre for Smart Infrastructure and Construction

    Brian Sheil

    City-scale digital twins, fibre-optic sensing and data-driven asset management.

    Engineering (Civil)Engineering

  • Institute for Manufacturing

    Alexandra Brintrup, Ajith Parlikad

    The Supply Chain AI Lab and digital twins for asset management.

    Engineering (Manufacturing and Management)Engineering

  • Alexei Lapkin

    Machine learning for reaction and process design. Co-leads the iDMT innovation centre.

    Chemical Engineering and BiotechnologyChemistryEngineering

  • Ross King

    Science automation and the Robot Scientists Adam, Eve and Genesis.

    Chemical Engineering and BiotechnologyChemistryBiologyCross-cutting

  • Markus Kraft

    The World Avatar, a knowledge-graph digital twin.

    Chemical Engineering and BiotechnologyEngineeringChemistry

  • Theory Research Interest Group

    Michaelides, Collepardo-Guevara, Colwell, Goodman, Wales, Thom, Reinhardt

    Interfaces and catalysis, chromatin and condensates, protein function, chemical informatics and energy landscapes.

    ChemistryChemistryBiology

  • Lennard-Jones Centre

    Materials and molecular modelling, with an open discussion group on Mondays at 2pm.

    Several departmentsChemistryMaterialsPhysics

  • Materials Theory Group

    Chris Pickard, Bartomeu Monserrat

    Ab initio random structure searching and data-derived potentials.

    Materials Science and MetallurgyMaterials

  • AI for Materials

    Jacqueline Cole

    Physics-informed AI, data-driven materials discovery and ChemDataExtractor.

    Physics (Cavendish)MaterialsPhysics

  • Electron Microscopy Group

    Paul Midgley, Caterina Ducati

    Advanced data analysis for electron microscopy.

    Materials Science and MetallurgyMaterials

  • CMBAgent Lab

    Boris Bolliet

    Multi-agent AI for cosmology: cmbagent and Denario.

    Physics (Data Intensive Science)PhysicsCross-cutting

  • Cosmology and astrostatistics

    Will Handley, Hiranya Peiris, Kaisey Mandel, Vasily Belokurov, Anastasia Fialkov

    Nested sampling, Bayesian machine learning, astrostatistics, Gaia and 21-cm cosmology.

    Cavendish, Institute of Astronomy and the Kavli InstitutePhysics

  • Astro Automata

    Miles Cranmer

    Symbolic regression with PySR and the Cambridge node of Polymathic AI.

    DAMTP and the Institute of AstronomyPhysicsCross-cutting

  • Cambridge Image Analysis

    Carola-Bibiane Schönlieb

    Inverse problems, scientific machine learning, biomedical imaging and the mathematics of deep learning.

    DAMTPCross-cuttingBiologyPhysics

  • van der Schaar Lab and CCAIM

    Mihaela van der Schaar

    Machine learning for medicine. Lab PhD applications are open for 2027 entry.

    DAMTPBiologyCross-cutting

  • Statistical Laboratory

    Samworth, Shah, Bacallado, Zhao, Nickl, Loh, Jog, Aston, Mandel

    High-dimensional statistics, causal inference and statistical learning theory.

    DPMMSCross-cutting

  • Machine Learning and AI, Computer Science

    Lawrence, Liò, Jamnik, Huszár, Lane, Mishra

    Machine learning and the physical world, graph networks for multi-omics, AI for mathematics and machine-learning systems.

    Computer Science and TechnologyCross-cuttingBiology

  • Teichmann Lab

    Sarah Teichmann

    Single-cell and spatial genomics with machine learning, and the Human Cell Atlas.

    Cambridge Stem Cell Institute and Department of MedicineBiology

  • Wellcome Sanger Institute

    Cellular Genomics and Generative Genomics programmes.

    HinxtonBiology

  • EMBL-EBI

    Host of the AlphaFold database, with research groups in protein families, genomics and computational biology.

    HinxtonBiology

  • Scheres group, MRC Laboratory of Molecular Biology

    Sjors Scheres

    Cryo-EM reconstruction methods, including RELION.

    MRC LMBBiology

  • Markowetz Group

    Florian Markowetz

    Computational oncology.

    CRUK Cambridge InstituteBiology

  • Milner Therapeutics Institute

    Namshik Han

    AI for drug discovery, and the monthly AI Club for Biomedicine.

    Cambridge Biomedical CampusBiologyChemistry

  • MRC Biostatistics Unit

    Biostatistical machine learning among five research themes.

    School of Clinical MedicineBiologyCross-cutting

  • Earth Sciences and the British Antarctic Survey

    Gualandi, Mashayek, Branson

    Environmental risk and ocean and ice modelling.

    Earth Sciences; BASEngineeringPhysics

  • Institute of Computing for Climate Science

    Emily Shuckburgh

    Climate computing and research software such as FTorch, with a summer school and paid internships.

    DAMTP, Computer Science and Cambridge ZeroEngineeringCross-cutting

Group pages change often. If a link is out of date or a group is missing, tell us.

§3Start this week

You do not need to be admitted to anything to start.

Open to anyone at Cambridge

Find a supervisor in four steps

  1. Pick two groups from the directory and read one recent paper from each.
  2. Email the principal investigator a one-paragraph fit statement and your CV: which project, and why you.
  3. Ask about a named slot, such as a Part II, Part III or MPhil project, or a UROP placement. A specific opportunity is easier to say yes to than an interest in AI.
  4. Apply for an Engineering UROP: paid at or above the National Living Wage, about ten weeks in summer, and open to projects outside your own department.

Engineering UROP details

No machine-learning background? Start the self-paced Accelerate Python for Science and AI Core Concepts modules this week, book one Accelerate workshop, then come to a Monday Lennard-Jones discussion group and ask a question.