I design evidence-first AI workflows, agent skills, and verification systems that help teams distinguish real completion from confident claims. My work focuses on AI agent QA, prompt engineering, delivery verification, acceptance criteria, runtime evidence, contradiction detection, recovery planning, and safe human approval gates. I build practical tools for founders, agencies, project managers, consultants, and teams working with AI agents, freelancers, employees, and vendors. Core principle: “Done” is not accepted until the required output, tests, read-back, deployment, approval, and recovery evidence match the original task. My products are designed to be structured, reusable, fail-closed, and useful in real business workflows.