Archives
All the articles I've archived.
-
AI Evals in Healthcare
Evaluating AI applications is necessary for prod deployments. It's also hard. AI in Healthcare demands a higher bar for safety and reliability. This post talks about how AI Evaluations ought to be implemented for safe deployment in Healthcare.
-
AI Evaluation in Practice
Team roles, responsibilities, and practical considerations for implementing AI evaluation systems in production healthcare environments.
-
Stages of Evaluation
How to approach AI evaluation at different product maturity stages - from early demos to production deployments.
-
Ideal Safety Spec
Essential components every AI app deployed in healthcare should have for safety and reliability.
-
AI Evaluation Framework
Hamel's observe-judge-improve framework for evaluating AI systems, from telemetry to domain experts to LLM-as-judges.
-
AI Safety is a Hard Problem
Building AI demos is fast; building production-safe systems is exponentially hard. An introduction to evaluating AI deployments in healthcare.