Build a systematic evaluation framework for a Retrieval-Augmented Generation pipeline that measures retrieval accuracy, answer faithfulness, and response relevance using automated LLM-as-judge scoring.
All ML Pipeline Design agents · All Data Science & AI agents