The RAG and Vector Search Roadmap

What retrieval-augmented generation is, why models need it, when not to use it, your first real embeddings, and the product this course builds step by step.

Keyword search by hand: terms, the inverted index, TF-IDF and BM25, phrases and fields, real engines, and the baseline every later RAG idea must beat.

Turn text into vectors, compare them with cosine, store them, call embedding models safely, and build AskDocs' embedding step with a cache.

Search a million vectors fast: exact search first, then IVF and HNSW built by hand, smaller vectors, recall measured honestly, and filters.

Load Markdown, HTML, PDFs and tables, clean and de-duplicate them, chunk with context and metadata, and re-ingest only what changed.

Store embeddings so they survive restarts, take updates, answer filtered queries and fit in memory: pgvector, other stores, sizing and a VectorStore.

Run keyword and vector search together, fuse them with reciprocal rank fusion, re-rank a shortlist, keep results diverse, and measure whether each stage pays.

The question is the weak link: rewrite it, expand it, split it, resolve "it" in a conversation, route it, and read filters out of it before retrieval runs.

Turn retrieved passages into an answer you can trust: prompt layout, a context budget, checked citations, a calibrated refusal and conflicting sources.

Golden sets, recall, MRR and nDCG by hand, answer checks and judges, failure triage, regression tests in CI and live signals, ending in an eval harness for AskDocs.

Keep untrusted document text from giving orders, keep private text private, keep tenants apart, and delete for real: the defensive side of RAG.

From a notebook that works to a service that keeps working: two paths, freshness, caches, budgets, cost, traces, scaling, model changes and incident drills.

Parent-child retrieval, summaries, multi-vector, graphs, SQL, images and search as a tool: when plain chunks and top-k are not enough, and what each costs.

Turn twelve phases into interview answers and one shipped product: design questions, debugging method, trade-offs, and AskDocs assembled end to end.