At TapestryHealth, I designed and built an agentic QA harness that analyzes requirements, generates and runs tests, triages failures, and produces evidence-backed results. I also built an LLM-as-judge parity harness to score agent and API outputs before release, with HIPAA-grade safeguards to keep evaluation runs from exposing PHI.
I created Verdict, an open-source evaluation harness for AI agents, with injection-resistant judging, HIPAA and NIST AI RMF compliance modules, and CI/CD quality gates. Across regulated identity and clinical communication products, I built test automation and validation workflows, including biometric enrollment and verification suites and checks for backend data integrity.

