Pillar 3

RAG (Basic → Advanced → Eval)

build RAG that works in production. Biggest hidden gap - basic tutorials stop at "embed + top-k," real systems need hybrid search, reranking, evaluation

3a · Foundations

Why RAG; embeddings; vector DB; cosine KEY
RAG from First Principles - Ankit Bansalunverified38m⚠ link unverified
Build a real RAG app
Ed Donner - LLM Engineering, Wk5 (knowledge-worker)repoUdemy

3b · Advanced retrieval (the gap)

Hybrid search: BM25 + dense + RRF
Production RAG w/ LangChain & Vector DBs - freeCodeCampvideo7h38m
Reranking (cross-encoders/ColBERT)
Advanced RAG techniques - Field Guide to AIarticle9m read
Query transforms (HyDE, multi-query)
Advanced RAG: hybrid, reranking, query expansionarticleread
Chunking: parent-child, sentence-window
12 Advanced RAG Techniques (2026) - Atlanarticleread
GraphRAG / RAPTOR / Contextual Retrieval
Same Atlan guide (techniques 8-12)articleread

3c · RAG evaluation (RAGAS) [KEY]

Metrics: faithfulness, context precision/recall
RAG Evaluation, RAGAS explained - Logical Lensesvideo12m
RAGAS hands-on in Python
RAG Evaluation Metrics Tutorial - Analytics Vidhyaarticle6m