index / evals
evals
public demos of how i score models on real work. each one is a write-up plus the data behind it.
- jev vs llms, closed and opena decision model against Claude and three fine-tuned open models on hand-labeled Texas RFPs.sep 2026
- expert capture: payroll opsa payroll workflow, a rubric, and a 3×3 eval across three Claude models.oct 2026