Skip to main content

Topic roadmap

Retrieval and RAG

15 lessons, about 49 hours in all, from the basics to advanced. Start with the first, or jump to what you need.

Lessons, easiest first

  1. Retrieval and RAG · Intermediate 2 · about 6 hBuild a RAG appBuild question answering over your own documents that retrieves first, then answers with sources.
  2. Retrieval and RAG · Intermediate 2 · about 3 hCheck whether retrieval worksMeasure retrieval with a small test set, using metrics such as recall at k and context precision.
  3. Retrieval and RAG · Intermediate 2 · about 2 hChunk documents for retrievalSplit documents into chunks that retrieval can find, and choose a chunk size for a task.
  4. Retrieval and RAG · Intermediate 2 · about 3 hEmbeddingsExplain what an embedding is and use one to find which sentences mean similar things.
  5. Retrieval and RAG · Intermediate 2 · about 4 hEnterprise connectorsPull documents and records from SharePoint, Salesforce, SAP, Jira and cloud storage reliably: authentication, paging, rate limits, incremental sync, deletions and permissions.
  6. Retrieval and RAG · Intermediate 2 · about 3 hGrounding, citations and refusing to guessMake a RAG system answer only from its sources, cite the passage behind each claim, and say plainly when the sources do not hold the answer.
  7. Retrieval and RAG · Intermediate 2 · about 4 hHybrid search, reranking and filtersCombine keyword and vector search, rerank the merged results with a cross-encoder and narrow searches with metadata filters.
  8. Retrieval and RAG · Intermediate 2 · about 2 hLong context or RAGDecide when to put whole documents into a long context window and when to retrieve, weighing accuracy, cost, latency and the number of documents.
  9. Retrieval and RAG · Intermediate 2 · about 2 hMetadata that helps retrievalStore the metadata that makes retrieval precise (source, section path, dates, type, owner, language, access) and add generated metadata such as summaries and likely questions where it helps.
  10. Retrieval and RAG · Intermediate 2 · about 3 hQuery rewriting, HyDE and parent-child chunksFix the questions plain retrieval gets wrong: rewrite vague or compound queries, search with a hypothetical answer, and return whole sections while matching small chunks.
  11. Retrieval and RAG · Intermediate 2 · about 3 hVector searchStore embeddings in an index and retrieve the closest chunks for a question.
  12. Retrieval and RAG · Intermediate 3 · about 3 hLong documents without vectorsRetrieve from long, structured documents by having a model navigate their table of contents, and know when this beats vector search.
  13. Retrieval and RAG · Pro 1 · about 4 hGraph RAGKnow when a knowledge graph improves retrieval, how GraphRAG builds entities, relationships and community summaries from documents, and what it costs compared with ordinary RAG.
  14. Retrieval and RAG · Pro 1 · about 3 hMultimodal RAGRetrieve and answer over images, charts, tables, slides and scanned pages, either by converting them to text or by retrieving page images and answering with a vision-language model.
  15. Retrieval and RAG · Pro 1 · about 4 hPermission-aware RAGMake a RAG system return only passages the current user may see, by storing access rights with each chunk, filtering inside the search and keeping rights in sync with the source.

On these role roadmaps

Interview questions on retrieval and rag

Books on retrieval and rag

All recommended books

Today's AI turns are used

Lessons, roadmaps, videos, practice cards and the code playground stay free with your account, with no limit.