Work
Built and shipped.
Two public apps you can try, a peer-reviewed ML paper, a talk, and a book.
Clinical Protocol Analyzer
An agentic tool that reads a clinical trial protocol, maps every section to the ICH-GCP E6(R2) checklist, and flags what is missing, with the clause cited for each gap. Paste your own excerpt for a live review. I designed the pipeline, wired the model, and shipped the app end to end.
Try the analyzerLive appDrug Repurposing Explorer
A knowledge-graph model that ranks repurposing candidates for approved drugs, each with the graph path that explains why. Public, no login. I designed the architecture, built the model, and shipped the app end to end.
Try the demoPeer-reviewedScientific Reports 2025
First-author paper: Self-supervised learning reduces label noise in sharp wave ripple classification. Methods proven on real biological data at scale.
Read the paperTalkDrug repurposing with AI
IAAM meetup, Marseille, July 2026. From knowledge graph to explainable prediction, demonstrated live. Slides and notebook to download.
Slides & notebookBookAI in Healthtech Without the Hype
A practical book for teams building healthcare AI: what works, what quietly breaks, and how to ship past the POC stage.
Read the book