Note 01 1 of 10
How RAG Works, from Documents to Answers
Understand how RAG connects documents to an LLM: build an index, retrieve passages, assemble context, and return an answer with traceable sources.
Collection
RAG Engineering: From Retrieval to Reliable Answers
Ten articles for software engineers on how RAG systems work and fail. Follow worked examples through parsing, indexing, retrieval, context assembly, grounded answers, evaluation, and security.
10 notes in this path
Note 02 2 of 10
Chunking, Parsing, and the Evidence You Lose
Compare chunk boundaries on the same source, preserve qualifiers and table structure, and evaluate retrieval against stable evidence rather than arbitrary chunk IDs.
Note 03 3 of 10
Versioned Ingestion and Safe Index Publication
Design document ingestion around immutable representations, idempotent jobs, and atomic publication so failed updates cannot silently replace searchable evidence.
Note 04 4 of 10
Embedding Models and Vector Index Compatibility
Understand vector geometry, pin the full encoding contract, and evaluate approximate search under real filters before migrating an embedding index.
Note 05 5 of 10
Context Assembly, Parent Retrieval, and Multimodal Evidence
Turn retrieved candidates into useful context by preserving qualifiers, tracking expansion provenance, budgeting tokens, and separating image retrieval from visual answering.
Note 06 6 of 10
Hybrid Search, RRF, and Bounded Corrective Retrieval
Work through reciprocal rank fusion, distinguish full-text search from BM25, and evaluate reranking and corrective retrieval without treating scores as confidence.
Note 07 7 of 10
Query Planning and Safe Structured Retrieval
Translate questions into bounded retrieval plans, preserve metadata constraints, and verify SQL result semantics with a worked incident-table example.
Note 08 8 of 10
Grounded Answers, Citation Validation, and Failure Recovery
Separate schema validity, citation identity, and factual support; design bounded repair, honest abstention, and streaming behavior around those distinct guarantees.
Note 09 9 of 10
Evaluate Retrieval and Gate Regressions
Separate retrieval coverage, context quality, and answer support. Calculate ranking metrics and design reproducible regression gates.
Note 10 10 of 10
Secure and Operate a RAG System
Carry authorization through retrieval, context, caches, and citations. Define trust boundaries, revocation behavior, and operational controls.