Topic overview
Study ingestion, chunking, embeddings, vector search, hybrid retrieval, reranking, prompt assembly, and citations.
Core concepts
Recall, precision, top-k, metadata filters, context compression, answer grounding, freshness, and evaluation datasets.
Why it matters
RAG helps AI systems answer with private, current, and source-backed information without retraining the model.
Interview relevance
Strong answers debug retrieval separately from generation and explain how quality will be measured.