(Senior) Postdoctoral Research Scientist – Generative AI and Closed-loop Discovery in Norwich

The post holder will conduct primary research in the AI for Biology Group to develop generative, causal and decision-making AI methods for biological design and closed-loop discovery. The role will focus on Bayesian decision-making, experimental design and original AI algorithms for modeling high-dimensional biological design spaces with interpretable and uncertainty-aware representations. Using these representations, the post holder will develop methods to generate testable hypotheses, propose candidate biological designs, prioritise experiments, reason over biological constraints and learn from experimental feedback. This post will form the design and discovery engine of the Generative Digital Biology programme.

Working alongside established foundation model research in the group, the successful candidate will develop methods that connect biological representation learning with generative modelling, reinforcement learning, Bayesian experimental design, active learning, uncertainty quantification, causal modelling, multi-objective optimisation and combinatorial optimisation. The successful candidate will join at a rare moment: early enough to help shape a new AI for Biology programme at EI, but with strong algorithmic foundations, prior publications, existing collaborations and a clear research trajectory already in place. The ambition is to move beyond models that only predict biological properties, towards AI systems that can propose, refine and prioritise biological designs and experiments.

Application areas may include sequence and RNA design, regulatory elements, perturbation design, synthetic constructs, cellular states, genotype-to-phenotype landscapes, engineering biology, plant systems, human health and therapeutic discovery collaborations where appropriate. This will be a highly collaborative role embedded across EI. The post holder will work with EI colleagues and platforms to connect AI-designed hypotheses and candidates with biological data, experimental design and validation routes, including potential collaborations with Earlham Biofoundry and engineering biology colleagues on AI-guided design-build-test-learn cycles; with Cellular Genomics and Single-cell and Spatial Analysis on perturbation, cell-state and single-cell/spatial omics use cases; and with BioFAIR, ELIXIR-UK, Open and FAIR Data and Research e-Infrastructure colleagues on AI-ready design datasets, benchmarks, provenance and reproducible workflows.

The post holder will be expected to lead high-quality research outputs, publish in leading AI, machine learning, computational biology and life science venues, contribute to open and reproducible algorithms, software and benchmarks, and support future competitive grant applications to UKRI, EPSRC, BBSRC, Wellcome, ERC and related funders.

Key Relationships

INTERNAL: Reporting to Professor Ke Li, the post holder will work closely with the AI for Biology Group and collaborate across EI’s research programmes, National Bioscience Research Infrastructures and technology platforms. Key internal relationships are expected to include BioFAIR, ELIXIR-UK, and Open and FAIR Data colleagues; Research e-Infrastructure; the Cellular Genomics programme; the Single-cell and Spatial Analysis platform; Earlham Biofoundry and engineering biology colleagues; Transformative Genomics; High-Performance Sequencing; and relevant EI scientific groups working on plants, microbes, biodiversity, health, genomics and data-intensive bioscience. The role is intended to help make the AI for Biology Group a collaborative AI engine for EI, supporting AI-ready design datasets, generative-design benchmarks, provenance-aware experimental records, model-guided experimental design and closed-loop discovery workflows across the Institute. Internal and external collaborations may occur as described.

EXTERNAL: The post holder will interact with UK and international collaborators in AI, machine learning, computational biology, genomics, single-cell and spatial biology, engineering biology, plant science, human health and therapeutic discovery. External collaborations may include academic, clinical, public-sector, infrastructure and industry partners where appropriate.

Main Activities & Responsibilities

• Develop original generative, causal AI and optimisation methods for biological design, hypothesis generation, perturbation prioritisation and experimental discovery.
• For appointment at SC5, take intellectual and operational leadership of a defined generative or closed-loop discovery workstream, set scientific priorities and milestones, manage technical risks, and deliver the work with limited supervision (essential for SC5).
• Develop theoretical foundations and practical algorithms using approaches such as diffusion models, flow models, autoregressive models, energy-based models, reinforcement learning, Bayesian optimisation, active learning, causal learning, multi-objective optimisation or combinatorial optimisation.
• Develop closed-loop experimental design methods that combine uncertainty quantification, multi-fidelity modelling, safe exploration, biological constraints and lab-in-the-loop feedback.
• Integrate foundation models, biological priors, mechanistic knowledge, causal representations, genotype-to-phenotype landscapes or fitness landscapes to guide the design of DNA/RNA/protein sequences and functions, regulatory elements, perturbations, synthetic constructs or cellular states.
• Collaborate with Earlham Biofoundry, engineering biology, Cellular Genomics, Single-cell and Spatial Analysis, BioFAIR, ELIXIR-UK and other EI colleagues to identify biological use cases, define AI-ready design datasets and prioritise candidates for experimental validation. For appointment at SC5, coordinate the relevant interdisciplinary collaboration and take responsibility for translating methods into a coherent experimental-validation plan (essential for SC5).
• Develop benchmark tasks, ablation studies, robustness/generalisation analyses, constraint-satisfaction evaluations, uncertainty estimates and biological validity checks for generative and design algorithms.
• Prepare manuscripts and conference papers for leading AI, machine learning, computational biology and life science venues; present findings internally, nationally and internationally. For appointment at SC5, lead the preparation and submission of major research outputs and represent the work in relevant external forums (essential for SC5).
• Contribute to research proposals, grant applications, open-source software, reproducible workflows, benchmark documentation and good research practice, including responsible data handling and reproducibility. For appointment at SC5, make substantive contributions to grant development and provide scientific or technical guidance to junior researchers or students (essential for SC5).
• As agreed with line manager, any other duties commensurate with the nature of the role.

Person Profile

Education & Qualifications

• PhD (awarded or expected within 6 months) in Computer Science, Machine Learning, Artificial Intelligence, Computational Biology, Mathematics, Statistics, Physics, Engineering or a related quantitative discipline (Essential).

Specialist Knowledge & Skills

• Strong knowledge in one or more of modern machine learning, generative modelling, reinforcement learning, Bayesian optimisation, active learning, uncertainty quantification, multi-objective optimisation or combinatorial optimisation (Essential).
• Excellent programming skills in Python and practical experience with PyTorch, JAX, TensorFlow, BoTorch, GPyTorch, Pyro, NumPy/SciPy or equivalent AI/scientific computing frameworks (Essential).
• Experience designing, implementing and evaluating original AI algorithms for design, optimisation, decision-making, experimental design, generative modelling or AI-for-science problems (Essential).
• Understanding of biological data or design problems, such as high-dimensional combinatorial or mixed-integer search spaces, DNA/RNA/protein sequences, genomics, transcriptomics, single-cell/spatial data, perturbation data, synthetic biology, engineering biology, molecular design or drug discovery (Desirable).
• Demonstrable experience in closed-loop or active experimental design, or an equivalent sequential decision-making setting with real-world feedback (essential for SC5) (Desirable).
• Experience with biological foundation models, sequence design, inverse design, structure-aware design, perturbation modelling, genotype-to-phenotype modelling or fitness landscapes (Desirable).
• A strong track record of independent or semi-independent research in machine learning, AI, computational biology, bioinformatics, or a closely related field (essential for SC5) (Desirable).
• Ability to develop and deliver a research direction with limited supervision, including project planning, collaboration and communication with interdisciplinary partners (essential for SC5) (Desirable).

Interpersonal & Communication Skills

• Ability to work independently, use initiative, solve complex research problems and deliver against agreed milestones (Essential).
• Excellent written and verbal communication skills, including the ability to communicate AI methods to biological collaborators (Essential).
• Ability to work collaboratively in an interdisciplinary team spanning AI, optimisation, computational biology, genomics, engineering biology and experimental biology (Essential).

Additional Requirements

• Attention to detail (Essential).
• Promotes equality and values diversity (Essential).
• Willingness to work outside standard working hours when required (Essential).
• Willingness to undertake occasional national or international travel for collaborations and conferences (Essential).
• Commitment to reproducible, open and responsible research (Essential).
• Willingness to embrace the expected values and behaviours of all staff at the Institute, ensuring it is a great place to work (Essential).
• Able to present a positive image of self and the Institute, promoting both the international reputation and public engagement aims of the Institute (Essential).
• Motivation to develop generative, causal and decision-making AI methods for biological design, experimental prioritisation and experimentally grounded closed-loop discovery (Essential).

About the Earlham Institute

The Earlham Institute harnesses the power of data-driven biology to accelerate solutions for health, biodiversity, and food security. Based at Norwich Research Park, the Earlham Institute is one of eight institutes strategically funded by BBSRC.

Our science combines world-class technology, interdisciplinary expertise, and training and development across genomics, engineering biology and data science, to decode the scale and complexity of living systems. We believe we can achieve more if we work together. That's why we collaborate with the global science community and industry partners, while also inspiring the next generation of scientists and technical specialists.

Our Science

Earlham Institute scientists specialise in developing and testing the latest tools and approaches needed to decode living systems and make biological predictions. We are home to state-of-the-art facilities and technology, creating a unique combination of expertise and infrastructure. We have dedicated laboratories for genome sequencing, single-cell analysis, engineering biology, and large-scale automation; as well as one of the largest supercomputing facilities for life science research in Europe. Our Advanced Training team also provides access to specialised scientific training to upskill the next generation of research and technical staff.

Our Culture

Our collegiate and innovative research environment comes with significant support, including a commitment to your professional development, research and administrative assistance, and opportunities to build collaborations with scientists and industry on the Norwich Research Park, across the UK, and internationally. The Institute is also home to talented technical and operational staff, whose invaluable contributions enable our science to have the maximum impact. We aim to recognise, reward, and develop all staff and students so that every individual feels able to achieve their best with us. We work hard to nurture an engaged and positive workplace, centred on core values that include openness, technical excellence, and collaboration. We attract staff from around the world who contribute to - and benefit from - an environment that enables them to deliver world-class science alongside a supportive and social community.

The AI for Biology Group will be newly established at the Earlham Institute (EI), while building on Professor Ke Li’s established and well-funded research programme in fundamental AI, biological foundation models and AI-driven scientific discovery. Recent work from the group spans foundation model pretraining and adaptation, genomic model evaluation and interpretation, RNA inverse design, and AI co-scientist systems. The group’s broader methodological foundations include data-driven, multi-objective and multi-fidelity optimisation, reinforcement learning, Bayesian decision-making and automated scientific discovery, with applications across RNA design, software engineering, renewable energy and other complex scientific domains. Further information is available at https://colalab.ai/.

At EI, the group will extend these foundations into a broader Generative Digital Biology (GDB) programme, developing multimodal, cross-organism and experimentally grounded AI systems that can learn from biological data, support hypothesis generation, guide experimental design and accelerate discovery across EI’s research programmes and technology platforms. The successful candidate will join a group with both ambition and infrastructure. The group is supported by substantial AI compute for frontier model development, including more than 70 dedicated GPUs, currently comprising 12 NVIDIA H200 and 8 NVIDIA H100 GPUs, alongside further capacity through the Norwich Data Centre and routes to national-scale AI compute such as Isambard-AI. This environment will support large-scale model pretraining, post-training, and adaptation, generative design, AI-agent systems and rigorous biological model evaluation. Access to and collaboration with the Earlham Biofoundry will create routes for model-guided experimental design and, where appropriate, closed-loop self-driving laboratory workflows supported by established lab automation.