Practiced one out loud? Paste your answer into the free self-check and get a score in 30 seconds
What the interviewer is testing
Whether you can explain the different outputs and jobs of the two model types and show how they combine in a retrieval system.
Answer framework
- Embedding model: maps a piece of text to a fixed-length vector so that similar meanings land close together.
- Generative model: predicts and outputs tokens to produce text, code or structured data.
- Uses of embeddings: semantic search, clustering, deduplication, classification and recommendations.
- Cost: embedding calls are far cheaper and faster, so you can embed a whole corpus once.
- Together: in RAG the embedding model finds candidate passages and the generative model reads them and writes the answer.
Sample answer
An embedding model turns text into a single vector of fixed length, for example a few hundred or a few thousand numbers. It is trained so that texts with similar meaning have vectors that are close by a measure such as cosine similarity. It does not write anything; its output is only useful for comparison.
A generative model produces tokens one at a time. It is what writes the answer, the summary or the SQL query.
They solve different problems and have very different costs. Embeddings are cheap enough to compute for millions of documents at ingestion, which is why they back semantic search, deduplication, clustering and simple classifiers. In a RAG system I embed every chunk once, embed each query at request time, find the nearest chunks, and only then pay for the generative model to read those few chunks and answer.
Likely follow-ups
Common mistakes
- Saying embeddings store the text; they store a lossy representation used for similarity.
- Mixing embeddings from two different models in one index.
- Using a generative model for tasks a cheap embedding classifier can handle.
- Forgetting to re-embed the corpus when the embedding model changes.
The 0 to 1 AI Engineer Interview Playbook covers the answer frameworks, system-design walkthroughs and an 8-week plan.
Get the playbook 40% off with code EARLYBF until Nov 3, 11:59 PM PTEARLYBF: 40% off until Nov 3Smaller start: Score your answers against 50 rubrics, $9. 50 AI engineer questions, one rubric each, PDF on Gumroad.
Interviewing for PM or TPM? See the full PM package, $59 on Gumroad: a resume system, a PM, TPM and PGM interview handbook, a 1:1 cheat sheet and a PM playbook. PM and TPM content, not AI engineer prep.
Related questions
Based on question #15 in the full question bank.