Your question is LLM Evaluation in Production. Take a moment with it on the right.
Talk me through your thinking if you like. When you're confident, submit your answer and I'll grade it like a real screen (7/10 or better passes).
How do you implement LLM evaluation frameworks to measure response quality in a live system?
Explain how you would define quality dimensions, build representative evaluation data, combine programmatic checks with LLM-as-judge and human review, and monitor quality after deployment. Include how you would detect hallucinations and prompt injection, calibrate evaluators, and balance evaluation coverage against latency and cost.
