Research Fellow in Machine Learning - Strand, London, WC2R 2LSAbout UsPharosAI offers a unique cancer AI product development ecosystem for drug discovery and clinical applications, democratising access to data, AI models, technologies, and capabilities.PharosAI unites large-scale multimodal cancer datasets with AI models through a highly-secure, trusted, federated platform, offering state-of-the-art AI tooling for use by pharma/biotech/life-sciences, AI-pharma, and AI developers of clinical applications, in enterprises, growth companies, and research organisations.PharosAI will revolutionise AI-powered cancer care, driving breakthrough therapies, clinical applications, addressing cancer’s social determinants, lowering barriers to UK SMEs, catalysing innovation and positioning the UK as a global leader in the dynamic healthcare and AI ecosystem. About The RoleOverview of the roleWe are seeking a highly motivated Machine Learning Research Scientist to join the PharosAI team. This role is focused on developing and implementing novel methods for federated learning in Cancer AI, with a particular emphasis on its application to multimodal data. The position will investigate open research questions and technical challenges in federated and decentralised AI for cancer research. This includes its application to modalities such as pathology, genomics, transcriptomics (spatial biology), proteomics, and electronic health records (EHRs). The role will also explore and address specific challenges arising from the multimodal nature of the data, such as data heterogeneity and hidden correlation structures, which can significantly impact both learning effectiveness and the design and robustness of privacy-preserving techniques. These techniques include, but are not limited to, differential privacy and homomorphic encryption. The candidate will contribute to the development of a federated learning platform within the PharosAI infrastructure, working closely with engineering teams. The ideal candidate will have a strong passion for research, software development and a meticulous approach to technical challenges, and the ability to work independently and collaboratively to drive projects to completion. The candidate will Lead the development of novel machine learning research methods in federated learning with a focus on multimodal cancer data, building on widely adopted algorithms such as FedAvg, FedProx, FedNova, Scaffold, and FedOpt to tackle challenges arising from the multimodal nature of cancer data. This includes data heterogeneity and hidden correlation structures, and their impact on learning performance. Implement and extend privacy-preserving techniques, particularly in the context of multimodal algorithms, e.g. differential privacy, secure aggregation and homomorphic encryption. Evaluate current opensource federated learning frameworks including NVFlare, Flower and OpenFL and contribute to their development with new methods for multimodal cancer data. Contribute to and support engineering teams in delivering a robust, production-ready federated platform for PharosAI. Support the engineering teams in the development and maintenance of MLOps pipelines to enable continuous integration, testing, deployment, and monitoring of machine learning models across the PharosAI platform. Develop and contribute to open-source projects aimed at the biomedical and healthcare research communities, extending toolkits such as MONAI to computational pathology and spatial biology. Publish research in high-impact, peer-reviewed conferences/journals. Present findings internally and externally at academic and professional conferences. Collaborate with a multidisciplinary team of engineers, data scientists, clinicians and biobankers. For Grade 8: managerial responsibility to supervise researchers (e.g. post-docs, PhD students) and lead research projects/programs. This is a full time (40 Hours per week) post, and you will be offered a fixed term contract until 31 March 2027