RAG Evaluation With RAGAS — Improve Retrieval Quality

I will evaluate your RAG system using the RAGAS framework — measuring faithfulness, answer relevancy, context precision, context recall, and context entity recall across a test set of representative questions — and delivering a quality improvement report identifying specific configuration changes targeting your weakest metrics. RAG systems deployed without structured evaluation have unknown performance characteristics — RAGAS metrics reveal whether retrieval or generation is the bottleneck, which determines whether you should improve chunking, retrieval parameters, prompt design, or the synthesis model.

The evaluation covers RAGAS test dataset generation from your knowledge base, metric calculation across 50+ test cases, per-metric analysis identifying failure patterns, configuration recommendations targeting each failing metric, and a re-evaluation after changes confirming improvement.