The emergence of Generative AI models as a multivariate and multi-modal projection onto an embedded space enables an new approach to Biomedical Research. Amongst other advantages, the embedding coordinates lend themselves to federated learning. The key challenge, however, is to pursue these new Agentic AI ecosystems without unscalable and privacy compromised downloads or installations. Instead, a P2P computational infrastructures are proposed, where multimodel biomarkers for cancer risk assessment can be developed.
A passion for Computational Biology as a re-usable construct that can inhabit agentic analytical data fabrics (see ARPA-H). Familiarity with web computing, and in particular with the JavaScript Web stack is the critical skillset. Training in quantitative biomedical sciences preferred.