I study what AI does
after the demo.
Independent work on evaluation, evidence, and applied AI systems.
Field Note / 004
Explainable AI is not the end state.
Understanding why an AI system behaves as it does is important. The ability to systematically improve that behavior is what ultimately matters.
Read Field Note 004Field Note / 003
Build Your PartyField Note / 002
The Benchmark Ends at DeploymentBuild Note / 002
The agent that had to ask first.
Homebase
Policy/Decision/Interceptor/Evaluation/SFT
The model learned to ask for permission. The boundary did not transfer reliably.
Read Build Note 002Build Note / 001
Separating Measurement From InterpretationTeaching Note / 003
Build the system. Then explain why it behaved that way.
CourseApplication Development with LLMsFocusEvidence, stopping rules, state, memory, decision logic
Read Teaching Note 003Teaching Note / 002
When Is the Evidence Enough?Teaching Note / 001
Constraint, State, Decision, Response