2026-08-30
A Python multi-agent decision system that connects intent understanding, clarification, candidate generation, risk review, and explanation agents with a deterministic ranking engine for stable, …
2026-08-19
Built an evidence-first Agent evaluation workbench with 60 samples across 10 task types, a five-dimensional rubric, A/B comparison, bad-case analysis, and quality, cost, P95 latency, and safety gates.
2026-08-09
A developing note on evaluating AI product opportunities by starting with user tasks, problem frequency, and success criteria before choosing where model capabilities belong.
2026-08-09
Problem it solves Most job-search prompts and skills stop at resume polishing or generic interview questions. This suite treats job search as a reusable operating system: build a candidate evidence …
2026-08-07
An evolving framework for examining AI products through six lenses: target users, core tasks, model capabilities, experience loops, business value, and durable advantages.
2026-01-01
Used structured specifications to organize requirements and interaction design, then worked with coding agents across frontend and backend implementation, integration testing, and launch coordination.