Agents that cooperate
Multi-agent systems that reason over real clinical context — orchestration, triage, and FHIR tools wired together with open protocols.
01 — About
Engineer by training, builder by instinct.
Hey, I'm Michael 
—
I've led large-scale transformations, started a company from scratch, and stayed close enough to the
code to keep contributing to open source. My early graduate work was about building datasets that taught
computers to do human-like tasks; that foundation still shapes how I think about AI today.
The through-line is translation: taking something genuinely hard — a FHIR server, a clinical workflow, a research model — and turning it into a product that a real person can actually use. That means fewer demos that go nowhere, and more systems in production where patients feel the difference.
If you're working on AI, healthcare, or just want a good conversation about where the two meet, grab some time with me. Career details live over on LinkedIn.
02 — Focus
Where I spend my time
Three overlapping problems I keep coming back to.
Clinical data, made legible
Interfaces and MCP servers that let care teams and patients interrogate health records in plain language instead of query syntax.
Product at scale
Taking AI initiatives from pilot to production across large health systems — driving adoption, cutting clinical workload, and owning the roadmap.
03 — Work
Projects & contributions
A mix of open source, research, and products built inside large health systems.
The Healthcare Agents Collective
A coordinated team of AI agents working together over Google's A2A protocol — orchestration, triage, and FHIR agents that collaborate autonomously to deliver better patient care.
MCP Server for FHIR
Query a FHIR server in plain language — e.g. "give me the list of patients called John Smith over age 55 taking diphenhydramine".
ReferWell
An open-source project that streamlines patient referrals between family doctors and specialists, improving coordination of care.
HospitalGPT
An exploration using AI agents (AutoGen, GPT-4, Mixtral) to manage simulated patient populations, demonstrating what AI could do for healthcare administration.
AI for Medical Imaging
Co-developed FibrosisNet for predicting pulmonary fibrosis, and led deployment of COVIDNet for COVID-19 detection with top pharma and hospitals including Boston Children's.
Privata.ai & Patient Data Network
Co-founded a startup building a SaaS platform for internal data breach detection, based on my open-source Patient Data Network for secure patient data sharing.
AI Products at Providence
Led AI product initiatives impacting millions of patients — reducing clinical workload and enhancing patient engagement, driving significant user growth and operationalizing product strategy.
OpenMRS Java SDK
Developed an SDK integrating the open-source OpenMRS electronic medical record with the Privata.ai platform, facilitating secure data interaction.
04 — Elsewhere
Find me around the web
Publications, code, writing, and the professional timeline.