- Building evaluation + safety infrastructure for healthcare conversational agents.
- Shipping local-first clinical tooling and agent workflows.
- OpenScribe - Open source AI medical scribe (local-first, self-hostable).
- ProspectiveAgent (private) - Patient agent capable of calling patients and screening for conditions.
- prechart (private) - Precomputed patient context + action memory layer.
- historian (private) - Patient context / memory system.
- historian-mobile (private) - Mobile variant.
- Clinical-trial-pre-screener (private) - Calls patients to assess clinical trial eligibility.
- clinicaltrialqualifier - Determines if patient qualifies for clinical trials.
- PatientMessageRouter (private) - Message routing for patient workflows.
- AIAdvocate (private) - Synthesizes clinical conversations into usable summaries.
- sequential-diagnosis-ai (private) - Sequential diagnostic reasoning experiments.
- GI-Board-Examination - Automated upload/extraction/analysis pipeline. Hit SOTA on GI Board exam. Published it.
- tox (private) - Ingredient scanning prototype.
- Anti-Facenet - Adversarial privacy work (2020).
- Mediband - IoT bracelet for prescription management ($15k funding).
- AWS Comprehend Hackathon - Mined IRS 990 PDFs -> database -> mapped outputs for social work routing.
- On-skin impedance + deep learning for NAFLD detection - Advanced Science (2024)
- Nanoengineered immunosensors for acute thrombosis prediction - Science Advances (2024)
- AI-directed light-sheet microscopy - (2024)
- Refractive-index corrected light-sheet microscopy - Advanced Science (2025)
- Vision-language + LLM performance in gastroenterology - (2024)
- Multimodal LLM board-exam performance - medRxiv (2024)
- Improving readability of HF patient education materials using GPT-4 - JMIR Cardio (2025)
- Cardiac amyloidosis patient materials readability - medRxiv (2025)
- HF institutional materials readability - medRxiv (2024)
- USocial - Co-founded digital event marketing company (
$100k+ revenue), sold business ($200k+ revenue).
languages: [TypeScript, Python, Swift, HTML/CSS]
platforms: [Node.js, Electron, Web]
ml: [LLMs, eval pipelines, applied clinical NLP]
infra: [Docker, Postgres]