AI is on. The round is NOT about prompting — it's about judgment over what the AI gives you.
~25 min · run it top to bottom
- 1. Decompose yourself4 min4:00
Break the problem into pieces and state your plan BEFORE prompting.
Interviewer grades: Do you drive the solution, or follow the AI?
Say it in your own words — but hit these: restate the problem · data model · implement the core myself
checking microphone…
- 2. Prompt narrowly3 min3:00
Ask the AI for a skeleton or a test plan only. Keep the core business logic yours.
Interviewer grades: AI as a tool for scaffolding/edge-cases, not a crutch for the thinking.
- 3. Read before you run3 min3:00
Read every line it produced and predict what it does before executing.
Interviewer grades: Do you trust output blindly, or verify it?
- 4. Verify + catch the bug10 min10:00
Run your own tests. Find what the AI got wrong (off-by-one, missing edge, wrong dedup) and fix it yourself.
Interviewer grades: THE skill of this round — spotting wrong AI output fast.
- 5. Own the decisions5 min5:00
Explain the design choices and tradeoffs as yours. You're the engineer; the AI typed.
Interviewer grades: Ownership and judgment over generated code.
Say it in your own words — but hit these: final judgment · trade-off · treat ai output as unverified
checking microphone…