MockDojo

Question bank, AI-assisted, AI engineering, concepts

Evaluate an LLM feature

You shipped an LLM feature that summarizes support tickets. How do you know it's good, and how do you keep it from getting worse when you change the prompt or model?

20 minutes, for junior, mid-level, senior engineers

How the interview runs

  1. 1.Explain. Start with the core idea. How would you explain it to a teammate?
  2. 2.Go deeper. Go one level deeper. What actually happens, and where does it break?
  3. 3.Trade-offs. What would you do in practice, and what are the trade-offs?

What you're graded on

  • Evals that catch regressions30 pts
  • Mechanism explained30 pts
  • Depth and failure modes20 pts
  • Practical trade-offs20 pts

Practice this one for real

Answer it phase by phase, then get the rubric score, what you missed and the model answer outline.

Unlock with Lifetime