Engineering tools
LLMCheck
I built LLMCheck to capture a failed model response, review the expected behavior, and rerun the case as a regression test. The offline example lets engineers inspect the verdict without an API key.
Read the case study · Run the offline example
Coding benchmarks
For a professional AI evaluation engagement, I built reproducible coding benchmarks with reference implementations and behavioral tests. The tests made the expected behavior explicit so reviewers could compare generated implementations against the same requirements.
career-agent
I built a workflow that prepares application materials and fills supported form fields. The applicant reviews the work and controls what gets submitted. Read the case study.