NLP and retrieval interview questions
Retrieval has become central to applied NLP, so these questions cover both the classic methods and the RAG systems built on them. A typical one: a support chatbot gives a confident, wrong answer even though the documents it retrieved were correct. Where would you look?
Expect questions on bag of words and TF-IDF, when keyword search beats vector search and when hybrid wins, what an embedding captures, how vector databases trade accuracy for speed, and where reranking fits. Interviewers want you to find which stage of the pipeline failed before reaching for a fix.
Try one now
A RAG-based support chatbot gives a confident, detailed, but factually wrong answer about a product feature. The retrieved documents were actually correct. What's the most likely cause, and how would you confirm it?
Free, no sign-up. The interviewer asks follow-ups on your reasoning, then rates your answer.
What the 16 questions cover
- Embeddings6
- Vector databases6
- RAG5
- Bag of words3
- TF-IDF2
- BM25 and keyword search2
- Keyword vs vector vs hybrid search2
- Reranking2
- Tokenisation1
- Named entity recognition1
- Topic modelling1
- Chunking strategies1
A question can cover more than one topic, so the counts add up to more than 16.
How the practice works
- The AI interviewer asks the question and follows up on your answer, the way an interviewer would.
- Stuck? Ask for a hint. Each one gives a little more away, so you can get unstuck without being handed the answer.
- Get a written debrief. Four ratings, each from Needs work to Interview-ready, tied to what you actually said, and what to study next.