RAG
Also written as Retrieval-Augmented Generation
A technique that lets an LLM pull in relevant information from an external source (like a company's documents) at the moment it answers, instead of relying only on what it memorized during training.
Think of it like
Like an open-book exam — instead of relying purely on what the model memorized while studying, it can flip to the actual textbook page before answering.
Junior or senior?
Junior sounds like
Called an API that already had retrieval wired up.
Senior sounds like
Has built the retrieval pipeline themselves — chunking, embeddings, search — and can walk through it.
Ask them
“Walk me through how you chunked and retrieved data for a RAG pipeline you built.”
Sounds like real experience
Describes an actual chunking or retrieval decision — chunk size, embedding model choice, a re-ranking step they added — and why it mattered for their results.
Probe further if
Describes RAG only at the definitional level — 'it looks things up before answering' — without describing a single decision they made about chunking, retrieval, or evaluating result quality.