0%
Preparing your experience
English 日本語
0 Start Learning

Your Bag

E-learning materials

Retrieval-Augmented Generation (RAG) Fundamentals

Ground AI answers in your own documents using chunking, embeddings, vector search, reranking, and citations that let users verify every response.

Material Overview
Retrieval-Augmented Generation (RAG) Fundamentals

Learn to build a retrieval pipeline that keeps a language model factual. This E-Learning Material covers document loading and chunking strategies, embedding models, vector databases, semantic and hybrid search, reranking, and prompt templates that force the model to cite its sources. You will finish with evaluation methods for retrieval quality and a tour of the failure modes that quietly produce confident but wrong answers.

Curriculum

Learning Path

Each level is enrolled separately. Work through them in order, or start at the level that matches your experience.

  1. Level 01

    Beginner

    Purpose

    Understand why retrieval exists by seeing what happens when a model answers from memory alone.

    Outcome

    You can explain what retrieval adds and run a basic question and answer flow over your own documents.

    What You'll Learn
    • Compare answers with and without supplied source material
    • Learn the parts of a retrieval pipeline and what each one does
    • Prepare a small document set and ask questions against it

    199 CREDS

  2. Level 02

    Intermediate

    Purpose

    Build the indexing side properly, because most retrieval failures start with how the documents were split.

    Outcome

    You can index a document set so the right passage is retrieved for a given question.

    What You'll Learn
    • Choose chunk size and overlap for the kind of documents you have
    • Select an embedding model and store vectors with useful metadata
    • Test retrieval on its own and inspect what comes back before generation

    378 CREDS

  3. Level 03

    Advanced

    Purpose

    Improve retrieval quality with hybrid search, reranking, and prompts that force the model to stay on source.

    Outcome

    You can raise answer quality measurably and trace every claim back to a passage.

    What You'll Learn
    • Combine keyword and semantic search to cover both exact terms and meaning
    • Add a reranking step and filter passages by relevance and recency
    • Write generation prompts that require citations and forbid unsupported claims

    793 CREDS

  4. Level 04

    Expert

    Purpose

    Evaluate and operate a retrieval system where correctness matters and documents change constantly.

    Outcome

    You can run a retrieval system in production and know exactly which stage to fix when answers degrade.

    What You'll Learn
    • Measure retrieval and answer quality with a labelled question set
    • Keep the index fresh with incremental updates and deletions
    • Diagnose failures across chunking, embedding, ranking, and generation separately

    978 CREDS