claude-hfi
Evaluated 100+ AI-generated backend and full-stack solutions against real-world engineering standards, flagging correctness, safety, and scalability issues in Python and Node.js codebases. • Assessed front-end behavior including state assumptions, data consistency, and integration boundaries affecting user trust and system reliability. • Compare multiple implementations and selected solutions based on trade-offs in performance, scalability, and risk, improving average solution quality scores by 20%. • Designed and validated prompts simulating attack vectors, edge cases, and misuse scenarios, improving robustness of AI-driven systems. • Applied structured reasoning around system failures, ambiguous inputs, and adversarial behavior, aligned with Trust & Risk Engineering principles.