RAG Implementation

Acknowledgement: Shared by Peter Bamuhigire, techguypeter.com, +256 784 464178.

Use When

• Use when implementing or upgrading a production RAG pipeline — query transformation, contextual compression, corrective retrieval, RAGAS evaluation in CI, multi-tenant isolation, cost decomposition, failure triage, and Anthropic-specific patterns.
• The task needs concrete code patterns, evaluation gates, and failure-mode rules rather than the conceptual "what is RAG" introduction.

Do Not Use When

• The reader still needs the conceptual model of RAG. Send them to ai-rag-patterns first.
• The task is embedding generation, chunking, or vector-store comparison primitives. Use vector-databases.
• The task is pgvector DDL or Postgres tuning. Use postgresql-patterns.
• The task is fine-tuning embedding models or LLMs.

Required Inputs