RAG & Retrieval Systems

Find it. Ground it. Cite it.

Chunking, hybrid retrieval, re-ranking, and grounded generation — build RAG that survives real corpora, real questions, and the index that went stale on Tuesday.

Intermediate4 modules · 12 lessons~118 min1060 XPAligned with DeepLearning.AI RAG + industry practiceStart with: Transformers & LLMs

Progress0/12

Module

Indexing: What the Model Gets to See

The offline half of the pipeline — chunking and embedding choices frozen at index time

RAG in One Pipeline8 min · 75 XPRAG in One Pipeline

Chunking: The Unglamorous Lever10 min · 85 XPChunking: The Unglamorous Lever

Picking an Embedding Model9 min · 85 XPPicking an Embedding Model

Module

Retrieval: Finding What's Actually Relevant

Sparse + dense fusion, ANN internals, and queries the user never quite wrote

Hybrid Retrieval: BM25 + Dense10 min · 90 XPHybrid Retrieval: BM25 + Dense

Inside the Vector DB11 min · 95 XPInside the Vector DB

Query Rewriting & HyDE9 min · 85 XPQuery Rewriting & HyDE

Re-Ranking: The Second Look10 min · 90 XPRe-Ranking: The Second Look

Module

Precision & Grounding

What goes in the context window, and making the model answer from it — not around it

Context Assembly: The Last 200 Tokens of Engineering9 min · 85 XPContext Assembly: The Last 200 Tokens of Engineering

Citations & Grounded Generation10 min · 90 XPCitations & Grounded Generation

The Failure-Mode Catalog10 min · 90 XPThe Failure-Mode Catalog

Module

Evaluation & Advanced Retrieval

Measure the system honestly, then reach for structure — graphs, coverage, and the questions vectors can't answer

Evaluating RAG: Retrieval Scores and Answer Scores11 min · 95 XPEvaluating RAG: Retrieval Scores and Answer Scores

GraphRAG & Structured Retrieval11 min · 95 XPGraphRAG & Structured Retrieval

Capstone project

Ask your docs

Build a mini RAG pipeline: pick a document collection (your notes, a textbook, docs), chunk + embed it, and answer questions through retrieval + generation.

Deliverable: A working notebook/demo plus a paragraph on where retrieval helps vs. hurts answers.

Milestones

Finish every lesson to earn the RAG & Retrieval Systems Certificate.