AI Systems · core
RAG
Retrieval-Augmented Generation: ground an LLM answer in retrieved context.
Mental model
RAG = retrieve relevant chunks, pack them into the prompt, then generate. The quality ceiling is set by retrieval — a perfect LLM cannot answer from context it never received. Most 'RAG is bad' problems are retrieval problems.
How to study RAG
Begin by restating the mental model in your own words, then connect it to a concrete system you have built or operated. Name the mechanism, the constraint it addresses, and the trade-off it introduces. Use Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks (Lewis et al.), Lilian Weng — How to Build an Open-Domain Question Answering System?, LangChain — RAG from scratch to check details, but close the source before writing your explanation. Retrieval is the learning step; rereading is only preparation.
Next, compare RAG with Chunking, Hybrid Search, Reranking, Context Packing. Ask what changes in correctness, latency, resource use, operability, and failure recovery. Complete Build a minimal RAG pipeline and preserve the command, input, output, and one failed attempt as evidence. Finish by explaining the idea without jargon to someone who has not studied the track.
Proof of understanding
- Explain the mechanism from first principles and identify the state it reads or changes.
- Give one situation where the concept is the right choice and one where it is not.
- Predict a realistic failure mode before running the drill, then compare the prediction with evidence.
- Connect the result to a roadmap or build artifact instead of treating the concept as isolated trivia.
Where it matters
Doc Q&A, support bots, code assistants, HighSignal's intelligence engine.
Common mistakes
- Blaming the LLM when retrieval failed
- No eval set, so retrieval changes are guesswork
- Stuffing too many chunks and burying the answer
- Skipping reranking and hybrid retrieval
Learn from primary sources
Practice and explain it back
Build a minimal RAG pipeline
Chunk a small document set, embed and index it, retrieve for a query, pack context, and generate a grounded answer.
Expected evidence: Answers grounded in retrieved chunks, with sources.
Open the interactive drill →Review prompts
- Why is retrieval the quality ceiling of a RAG system?
- When would you choose RAG over fine-tuning?
Build evidence
RAG pipeline v0
An end-to-end retrieve → pack → generate pipeline grounded in a document corpus.
- Ingest documents with a chunking strategy
- Retrieve relevant chunks for a query
- Generate an answer grounded in retrieved context
- Have at least 10 eval questions
RAG system design doc
An architecture doc for a production RAG system with an eval gate.
- Offline ingestion and online query pipelines defined
- Eval harness as part of the architecture
- Freshness/re-indexing strategy