Artificial Intelligence Jobs in London

Hadean · London

Lead a new AI-powered defence product team, line-managing AI/ML, simulation, infrastructure and web engineers while staying hands-on in TypeScript, React and Node. You own delivery end-to-end, hiring and people development, partnering with Product and technical leads, with hybrid working one day a week in Shoreditch.

University of West London · London · £18,000 pa

Develops an AI-enabled decision-support framework combining building-performance simulation, machine learning, GIS and remote sensing to optimise green roofs and façades for energy, carbon and thermal performance. Primarily computational, with no routine lab or fieldwork, and scope refined with the successful candidate.

King's College London · London · £45,031 – £52,514 pa

Designs, builds and evaluates AI-enabled evidence synthesis tools and pipelines for the METIUS programme, writing production-quality Python code and onboarding consortium partners. Runs methodological and evaluation studies on the reliability and usability of AI-assisted synthesis methods, publishing the results, and contributes to the Solace-AI project on climate-health emergencies. Works alongside software developers, data scientists and evidence synthesis specialists.

The Francis Crick Institute · London · £47,500 pa

Lead a computational project modelling how driver mutation clones interact with their microenvironment in chronic liver disease and liver cancer. Develop graph neural networks, transformers and generative AI on large-scale spatial transcriptomics, single-cell and imaging data, in a cross-institutional collaboration with the Wellcome Sanger Institute's Lotfollahi Lab.

University of Oxford · Oxford · £39,424 – £47,779 pa

Deliver AI safety research for the ARIADNE consortium, a Wellcome-funded generative AI system conducting clinical interviews to assess psychosis risk. Develop LLM supervisor modules that detect delusional reinforcement and inappropriate responses to suicidal symptoms, evaluating them in controlled testbeds with simulated patients and LLM-as-judge scoring calibrated against clinicians.