AI agents need the same discipline as clinical ML
Why evaluation loops, governance, human review, and product accountability matter just as much for agentic systems as they do for regulated AI.
Technical Blog
Field notes on clinical AI, MLOps, and agentic systems.
Production lessons from patient-safety models, analytics platforms, governance loops, and AI products that need to survive contact with real users, real risk, and real operational pressure.
Search loads a compact index of published writing.
Latest
Recent writing
The model registry is not the platform
A practical view of streaming features, retrospective evaluation, clinical event definitions, deployment gates, and operational monitoring.
Clinical AI does not earn trust in a notebook
How model validation, advisory cadence, explainability, and workflow design decide whether a patient-safety model becomes real infrastructure.