Frontend engineer at Mono (YC W21). Building clinical AI evaluation infrastructure.
I focus on the harder problem: making clinical AI behave reliably in deployment contexts that benchmarks don't test for. MD background taught me to debug systematically.
- ClinicalGuard — Open-source evaluation framework for clinical AI systems, grounded in Nigerian Standard Treatment Guidelines (NSTG 2022). Hybrid retrieval over 251 conditions, four-dimension LLM-as-judge scoring with required/expected split, deterministic safety engine, measured intra-judge variance. Phase 2 complete. View repo →
- Claude Architect — Multi-agent clinical research system for GLP-1 receptor agonist literature. Orchestrator/Researcher/Critic architecture, 30-case adversarial eval suite, hybrid RAG over real trial data. View repo →
🩺 M.D. from University of Ibadan (2025). Self-taught engineering while completing clinical rotations.
AI/LLM: Claude API, prompt engineering, evaluation methodology, agent orchestration
Frontend: TypeScript, React, Next.js
Backend: Node.js, SQLite, PostgreSQL