Agents and AI engineering interview questions
Agent questions are system design questions with a twist: the part making decisions can be talked into things. Design an agent that reads customer emails and can issue refunds of up to $200. How do you make sure it can’t be persuaded to give money away?
Expect to discuss where limits belong (in code, not in the prompt), how to orchestrate several steps, how tracing helps you debug a run that went wrong, and how cost and latency add up with every model call. Interviewers look for answers that assume the model will sometimes fail, and plan for it.
Try one now
You're asked to build an LLM agent that reads customer emails and can issue refunds of up to $200 on its own. How do you design it so it can't be talked into giving money away?
Free, no sign-up. The interviewer asks follow-ups on your reasoning, then rates your answer.
What the 11 questions cover
- Agents6
- Tracing and observability5
- Cost and latency of LLM systems4
- Tools vs skills2
- Orchestration2
- Prompt injection and guardrails2
- MCP1
A question can cover more than one topic, so the counts add up to more than 11.
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.