Retrieval Systems
Retrieval Engineering
Lexical, dense, approximate, hybrid, and reranked search with measurable evidence quality.
10 chapters · 4-6 hours reading · 3-4 hours labs · 5-8 hours project
Chapters
- 1Retrieval begins with a relevance contractMental model
- 2Model the corpus before choosing an indexData architecture
- 3Lexical retrieval and BM25 from first principlesFormal model
- 4Dense embeddings as learned retrieval geometryRepresentation
- 5Approximate nearest-neighbor indexesSystems architecture
- 6Hybrid candidate generation and rank fusionImplementation
- 7Reranking, late interaction, and result diversityRanking
- 8Implement a retrieval pipeline with explicit stagesFrom first principles
- 9Evaluate retrieval as a chain of evidenceEvaluation
- 10Secure and operate retrieval in productionOperations