LLM105 · Retrieval-Augmented Generation (RAG)

Categories: AI & Automation
Wishlist Share

About Course

Build retrieval-augmented generation systems that answer questions from your own documents with citations, and measure whether retrieval and answers are actually right.

Advance · 8 hours · 4 lessons, a lab, an assessed assignment and a portfolio piece

Who it is for

Developers who can work with LLM APIs and structured outputs and now need assistants grounded in company knowledge.

Prerequisites

LLM103 and LLM104, or equivalent experience with LLM APIs, prompts and Python.

What you will be able to do

  • Design an ingestion pipeline that parses, cleans, chunks and indexes business documents with metadata
  • Implement dense, keyword and hybrid retrieval with optional reranking and compare them using recall@k
  • Generate grounded answers with source citations and a clear fallback when the documents do not contain the answer
  • Evaluate a RAG system on a question set for retrieval quality, faithfulness and answer correctness

Lessons

  • Lesson 1 · RAG architecture and document ingestion
  • Lesson 2 · Embeddings, search and reranking
  • Lesson 3 · Grounded answers, citations and evaluation
  • Lesson 4 · Lab and assignment

Lab: A cited Q and A assistant over a company handbook

You will index a sample employee and customer-service handbook, build dense, keyword and hybrid retrievers, and generate answers that cite the exact sections used. You will then measure retrieval recall and answer faithfulness on a set of 25 questions with known answers.

Free tools: Google Colab, sentence-transformers, ChromaDB (local), rank-bm25, and the Gemini API free tier or a local model through Ollama

Assessed assignment

Build a RAG assistant for a small property-management company that answers tenant and staff questions from its lease templates, maintenance policies and FAQ. Document your chunking choices, retriever comparison and prompt. Report recall@5, the share of answers fully supported by citations, and how the system behaves on at least five questions whose answers are not in the documents. Finish with a one-page list of the three most important improvements you would make next.

Criterion Weight What excellent work looks like
Ingestion and chunking 20% Documents are cleaned and chunked with justified sizes and useful metadata; tables and headings are not broken carelessly.
Retrieval quality 30% At least two retrievers compared on recall@5 with a clear winner chosen for documented reasons.
Grounded generation 25% Answers cite sources, stay within the retrieved text and fall back correctly on unanswerable questions.
Evaluation and next steps 25% Faithfulness and correctness are measured on a realistic question set, with honest analysis of failures and prioritized improvements.

Portfolio piece

A working, evaluated RAG assistant with documented chunking, retrieval comparison and grounded, cited answers.

Self-paced, 12-month access. Questions are answered in writing through the student panel. A completion record is issued when you finish the course.

Next step

Continue with LLM106 · Building AI Agents with Tools · 8 hours · $56.00

Continue to the next course

Prefer the whole program? Program in AI Application Development & LLM Engineering · $378.00 (10% less than buying the courses separately)

Show More

What Will You Learn?

  • Design an ingestion pipeline that parses, cleans, chunks and indexes business documents with metadata
  • Implement dense, keyword and hybrid retrieval with optional reranking and compare them using recall@k
  • Generate grounded answers with source citations and a clear fallback when the documents do not contain the answer
  • Evaluate a RAG system on a question set for retrieval quality, faithfulness and answer correctness

Course Content

LLM105 · Retrieval-Augmented Generation (RAG) (8 h)
Build retrieval-augmented generation systems that answer questions from your own documents with citations, and measure whether retrieval and answers are actually right.

  • Lesson 1 · RAG architecture and document ingestion
  • Lesson 2 · Embeddings, search and reranking
  • Lesson 3 · Grounded answers, citations and evaluation
  • Lesson 4 · Lab and assignment
  • LLM105 Assessment · Retrieval-Augmented Generation (RAG)
Scroll to Top