A field guide to the RAG sprawl

What it does

This is a curated awesome-list that maps the entire RAG ecosystem: frameworks, vector databases, evaluation metrics, architecture patterns, and a growing menagerie of advanced techniques. It collects links to tutorials, papers, and implementations across the whole pipeline from document ingestion to production deployment.

The interesting bit

The list attempts something genuinely useful: distinguishing between “naive RAG” (basic retrieve-then-generate) and a half-dozen architectural variants like agentic, self-reflective, and graph-based RAG. It also surfaces less-hyped approaches such as cache-augmented generation and temporal retrieval, which skip the vector-database-everything trend.

Key highlights

• Covers 20+ frameworks including LangChain, LlamaIndex, Haystack, and newer Rust-based tools like Swiftide and Kreuzberg
• Catalogs advanced techniques: GraphRAG, multimodal RAG, video RAG, corrective RAG, and retrieval-augmented fine-tuning (RAFT)
• Includes production considerations and evaluation metrics, not just toy implementations
• Maintains a separate deep-dive on the Python ecosystem
• Links to runnable Jupyter notebooks for chunking strategies, hybrid search, reranking, and query transformations

Caveats

• The README is heavy on definitions and benefits, light on critical analysis of when each approach actually wins
• Some advanced techniques link to papers or Medium posts rather than working code
• No clear criteria for what gets included or why one framework beats another

Verdict

Worth bookmarking if you’re building or evaluating RAG systems and need a map of the territory. Skip it if you already have strong opinions about chunking strategies and just want to argue about them on Hacker News.