Evaluation operations
Paired testing, regression review, launch gates, reviewer queues, runbooks, and decision records built to survive handoffs.
Safeguards operations / policy judgment / clinical systems
High-stakes judgment, made operational.
I am a licensed mental health clinician and operations leader with 15+ years of experience turning risk, policy, and quality expectations into decisions people can execute and audit.
I am now applying that background to AI safety evaluation operations: launch readiness, policy enforcement workflows, reviewer calibration, escalation, and documentation that keeps severe findings visible.
What I Bring
The operating disciplines behind trustworthy evaluation work.
Policy enforcement
Experience translating complex behavioral health standards into operational evidence, review criteria, and escalation paths.
High-stakes judgment
Clinical risk, crisis assessment, sensitive content, regulatory deadlines, ambiguity, and appropriate escalation under pressure.
Technical fluency
A working SQL analysis pipeline and interactive dashboard, paired with honest limits and a production-minded roadmap.
The Through-Line
Safety work is a chain of judgment, not a single score.
In clinical care, utilization management, quality improvement, and parity work, I learned that a standard only matters when people can interpret it consistently, document why a decision was made, recognize the edge case, and escalate before harm compounds.
Model evaluations need the same connective tissue. My portfolio shows how I would organize that work while remaining candid about the distinction between transferable operating experience and production trust-and-safety experience.
Selected Evidence
Finished work, with assumptions and limitations visible.
01