Choose who you’re hiring. Try the interview yourself.
AI Engineering5 exercises
Build systems that hold up.
Go beyond a working prompt. Explore retrieval, tool use, voice and multi-agent systems under real failure conditions.
Trace an agent’s decisions
Debug unreliable tools
Validate the fix
Inside the interview
The confidently wrong RAG bot
A support bot answers questions from an 8-chunk knowledge base. QA reports it gives confident but wrong answers, and when asked something outside the knowledge base it makes something up instead of saying it doesn't know.
Input→ Process→ Output
Explain what the code does, diagnose the issue, then make and test your change.
Maya asks candidates to explain their reasoning, edit code and test their changes. Your team can review the conversation alongside their work. No tool can reliably detect or block every external AI assistant; we don’t claim that we can.
Get in touch
Who is your next hire?
Leave your work email. We’ll help you set up your team’s pilot.