I work on ML and AI for biology, health and pharmaceuticals.
PhD in computational biology (LSHTM), applying machine learning and genome-wide association methods to bacterial genomics. Industry experience at AstraZeneca and BenevolentAI. Currently applying LLMs and foundation models to problems in biotech and pharma, with a particular focus on validation and provenance.
rbcl-generative-biology Prompt-conditioned generation and LoRA fine-tuning of a genomic foundation model (Evo 2) on rbcL. Corpus built with species-level leakage control and stratified splits; reproducible evaluation harness with memorisation controls. Manuscript in preparation.
kegg-string-mcp An MCP server exposing KEGG, STRING and PubMed as model-callable tools, with an agentic annotation pipeline that validates every citation against what was actually retrieved.
ccrcc-target-triage Drug-blind, multi-omic triage of candidate drug targets in clear cell renal carcinoma, using public proteogenomic data. Runs its own recovery test against known positives and reports the result either way.
womens-health-ml Whether passive wearable data can predict menstrual cycle phase. Reports a negative finding rather than a flattering one.