Track 09: Retrieval-Augmented Generation (RAG)

Grounding an LLM's answers in your own documents instead of its training data alone. Builds a naive RAG pipeline early, then spends the rest of the track on everything that makes the difference between a demo and a production system: parsing real documents, hybrid retrieval, reranking, citations, and honest evaluation.

Modules

  1. Why RAG Exists
  2. Naive RAG End to End
  3. The Ingestion Pipeline
  4. Document Loaders and Parsing
  5. PDFs, Tables and Complex Documents
  6. The Retrieval Step
  7. Keyword Search and BM25
  8. Hybrid Search
  9. Query Rewriting and Expansion
  10. Reranking With Cross-Encoders
  11. Context Assembly and Prompt Construction
  12. Citations and Source Attribution
  13. Evaluating Retrieval Quality
  14. Evaluating Generation Quality
  15. RAGAS and Evaluation Frameworks
  16. Advanced RAG: GraphRAG and Agentic RAG
  17. RAG Failure Modes
  18. Capstone Project