Build RAG you can trust in production.

RAG is retrieval-augmented generation: an LLM that answers from your documents instead of from its training memory.

• Setting Up
  • How to Use This Book
  • The Problem RAG Solves
• Build a RAG
  • Anatomy of a RAG System
  • Your First RAG
  • Watching Your Naive RAG Fail
• Embeddings
  • What an Embedding Is
  • Choosing an Embedding Model
  • The Geometry of Meaning
• Documents
  • Why Documents Aren't Ready for Retrieval
  • Chunking, the Most Underrated Decision
  • Metadata, the Free Win Most People Skip
• Storage
  • From Lists to Indexes
  • Choosing a Vector Store
  • Updating an Index Without Breaking Production
• Retrieval
  • Pure Vector Retrieval Is Not Enough
  • Hybrid Retrieval Done Right
  • Query Transformation
  • Reranking, the 80/20 Win
  • Filtering Without Killing Recall
• Generation
  • Stuffing the Context Window
  • Prompting for Grounded Generation
  • Citations and Attribution
  • Streaming Without Losing Citations
• Evaluation
  • Why "It Looks Right" Is Not Evaluation
  • Building a Golden Dataset
  • Retrieval Metrics That Matter
  • Generation Metrics: Faithfulness, Relevance, Completeness
  • The Eval Loop
• Production
  • Latency Budgets
  • Cost Optimization
  • Observability
  • Multi-Tenancy and Access Control
  • Failure Modes and Graceful Degradation
• Advanced
  • Hierarchical and Multi-Hop Retrieval
  • GraphRAG and Knowledge Graphs
  • Self-RAG and Corrective RAG
  • Agentic RAG
  • Multi-Modal RAG
• Capstone
  • The Capstone Project

$39 USD