Vertex

Retrieval-Augmented Generation from Scratch

Build a RAG system that actually retrieves the right thing: embeddings, chunking, hybrid search, reranking, and grounded answers.

Advanced1h 57m4 modules9.1k students

Start learning

What you’ll learn

  • Understand embeddings

    Know what a vector encodes, what it does not, and which similarity measure to use.

  • Chunk content well

    Split documents so retrieved passages are self-contained and carry useful metadata.

  • Improve retrieval quality

    Combine keyword and vector search, then rerank, and measure whether it helped.

  • Ground every answer

    Cite sources, and make the system say it does not know instead of inventing one.

Tracklist

4 modules · 1h 57m

  1. 1Embeddings and Vector SearchThe retrieval primitive: turning text into vectors and finding the near ones quickly.20m 54s
    1. Lesson 1.1What an embedding actually encodesFree preview8m 38s
    2. Lesson 1.2Vector databases and indexes4m 23s
    3. Lesson 1.3Similarity metrics and thresholds7m 53s
  2. 2Ingestion and IndexingGetting your content into the index in a shape that retrieves well and stays current.32m 24s
    1. Lesson 2.1Chunking strategies14m 2s
    2. Lesson 2.2Metadata and filtered retrieval12m 24s
    3. Lesson 2.3Keeping the index fresh5m 58s
  3. 3Retrieval QualityWhere most RAG systems are won or lost: combining retrieval methods, reranking, and measuring the result.24m 21s
    1. Lesson 3.1Hybrid keyword and vector search6m 47s
    2. Lesson 3.2Reranking retrieved candidates5m 15s
    3. Lesson 3.3Evaluating retrieval12m 19s
  4. 4Grounded GenerationTurning retrieved passages into an answer that is traceable, honest, and measurable.39m 32s
    1. Lesson 4.1Grounding answers with citations7m 16s
    2. Lesson 4.2Handling the no-answer case21m 33s
    3. Lesson 4.3End-to-end RAG evaluation10m 43s