Command Palette

Search for a command to run...

Hectal

RAG and Vector Search · from word matching to a production document Q&A service

Build RAG that answers from your documents.
One product, built phase by phase.

You need the basics of one programming language and a rough idea of what a language model is. The course builds one product, AskDocs, a service that answers questions from a company handbook with sources and refuses when the handbook has no answer, and every phase adds one production-grade part: keyword search, embeddings, vector indexes, ingestion, vector databases, hybrid search and re-ranking, query understanding, grounded answers, evaluation, security and operations. Every idea starts from why it exists and the simplest version built by hand. Every key example runs on the page in Python, Java and JavaScript with the same output, and search labs run a real embedding model in your browser so you can see real similarity scores.

Start from zero The full roadmap
Phases
14
Topics
130
Hours
50
Diagrams
132
Exercises
799

By the end you can

What this course makes you able to do

  • Decide when RAG is the right tool, and when a long prompt or fine-tuning is better
  • Build keyword search (an inverted index and BM25) and vector search (embeddings, IVF and HNSW) by hand, then use real search engines and vector databases
  • Ingest real documents: clean, chunk with context, attach metadata and re-index only what changed
  • Combine keyword and vector search, re-rank, rewrite questions, and write cited answers that refuse without evidence
  • Measure retrieval and answers with recall, MRR and nDCG, and keep RAG secure: access control, tenant isolation, prompt injection from documents
  • Run RAG in production, design RAG systems in interviews, and ship a complete document Q&A service

Beginner → Advanced

The curriculum, by level

Beginner
  1. 0Start Here: What RAG Is and What You Will Build8 topics · ~3 h
  2. 1Search Before Vectors9 topics · ~4 h
  3. 2Embeddings in Depth10 topics · ~4 h
Intermediate
  1. 3Vector Search: Exact, IVF, HNSW and Smaller Vectors9 topics · ~4 h
  2. 4Ingestion: From Files to Searchable Chunks10 topics · ~4 h
  3. 5Vector Databases10 topics · ~3 h
  4. 6Hybrid Search and Re-ranking9 topics · ~4 h
  5. 7Query Understanding9 topics · ~4 h
  6. 8Generation with Sources9 topics · ~3 h
Advanced
  1. 9Evaluating RAG10 topics · ~4 h
  2. 10Security and Privacy (Defensive)9 topics · ~3 h
  3. 11RAG in Production11 topics · ~4 h
  4. 12Advanced Patterns10 topics · ~3 h
  5. 13Interviews and Capstone7 topics · ~3 h

Where should I start?

Pick a path for your goal

New to RAG

You can code a little and want to understand retrieval from the ground up. Start at the first topic: every term is explained.

  1. 10 · Start Here: What RAG Is and What You Will Build
  2. 21 · Search Before Vectors
  3. 32 · Embeddings in Depth
  4. 43 · Vector Search: Exact, IVF, HNSW and Smaller Vectors

Shipping a RAG feature

You need document Q&A that works on real data: ingestion, a vector store, hybrid search and cited answers.

  1. 10 · Start Here: What RAG Is and What You Will Build
  2. 24 · Ingestion: From Files to Searchable Chunks
  3. 35 · Vector Databases
  4. 46 · Hybrid Search and Re-ranking
  5. 58 · Generation with Sources
  6. 69 · Evaluating RAG

Senior and interviews

You have built RAG and want evaluation, security, operations, advanced patterns and system design depth.

  1. 19 · Evaluating RAG
  2. 210 · Security and Privacy (Defensive)
  3. 311 · RAG in Production
  4. 412 · Advanced Patterns
  5. 513 · Interviews and Capstone