Skip to content
TariahAcademy
← Catalog

RAG · Builder

Retrieval-Augmented Generation (RAG)

Ground AI answers in your own documents with chunking, embeddings and evaluation.

Lessons

  • 1. Why models need retrieval

    Knowledge cut-offs, private data and citations.

    25m

  • 2. Chunking and embeddings

    Chunk size trade-offs and vector similarity.

    45m

  • 3. Retrieval and re-ranking

    Hybrid search and top-k selection.

    40m

  • 4. Evaluating RAG

    Faithfulness, relevance and test sets.

    35m

Exercises

Answers save in this browser. Compare yours with the checklist — no automated grading yet.

Chunk a policy

Split a 2-page HR policy into chunks. What size and overlap do you choose and why?

Write eval questions

Write 3 questions with known answers to test your RAG bot.