RAG Optimization

Techniques to improve every stage of the RAG pipeline — from smarter indexing and chunking strategies to hybrid search, re-ranking, and knowledge graph retrieval.

Topics

• Advanced Indexing — Moving beyond fixed-size chunking: parent-child indexing, summary indexing, and scalable vector storage strategies

• Hybrid Search — Combining dense (vector) and sparse (BM25) retrieval with Reciprocal Rank Fusion for more robust search results

• Query Transformation — Techniques to improve retrieval input: HyDE, query decomposition, step-back prompting, and query routing

• Post-Retrieval Processing — Re-ranking with Cross-Encoders, diversifying results with MMR, contextual compression

• GraphRAG — Building RAG systems that combine Neo4j graph databases with vector retrieval for richer reasoning over entities and relationships

• Multimodal RAG — Retrieval-Augmented Generation across text, images, and visual documents using vision-language models

Prerequisites

Complete Foundations first.

• Advanced Indexing
• Advanced Retrieval Strategies
• Query Transformations
• Re-ranking
• GraphRAG Implementation
• Multimodal RAG NEW
• Practice — RAG Optimization
• Assessment — RAG Optimization