Skip to content
AI Security

Interview question · Software, Data & Model Supply Chain

Who should be allowed to promote a model in the registry, and how do you detect tampering?

Promotion to production is a deployment decision dressed as a metadata change. The scored answer gives it change control, binds it to evidence, and makes swaps detectable.

mediumNewPreparation tags: Databricks, Google, AmazonUpdated Nov 2026

Promotion to production is a deployment decision dressed as a metadata change. The scored answer gives it change control, binds it to evidence, and makes swaps detectable.

Make your first attempt

What would you say before reading the answer?

  1. 01 State the assumptions your answer depends on.
  2. 02 Explain the mechanism and choose a control.
  3. 03 Name a condition that would change your decision.

The worked answer includes an approach, scoring criteria and follow-up questions. Compare the reasoning with your own.

10 answers per topic instead of 5, plus saved progress and bookmarks · costs nothingor unlock all 529 remaining answers · $25

Every worked answer in all 17 topics, every lesson in every course, the full concept curriculum, the complete practice-test bank, and the downloadable security field guides.

6 months · One payment · No auto-renewal

What if · the follow-ups in this answer

The answer works through 6 follow-ups like an interviewer would

Models move from a staging registry to production through a 'promote' button that any ML engineer can press, and the serving system loads whatever the production alias points to.

  1. Why is promotion a production change?
  2. How should duties be separated?
  3. What should the automated gates check?
  4. How do you detect tampering?
  5. What if an urgent rollback is needed?
  6. How would you redesign promotion?

Each rung has a nudge, a model answer, the principle that carries over and the weak answer to avoid. Premium opens every answer and ladder.

UP NEXT ON YOUR JOURNEY
Next in this trackWhat scans and admission policies would you require before an open-weights model runs in production?Next in this trackWrite a check that verifies downloaded model and dependency files against pinned hashes before they are used.Next in this trackHow would you vet a third-party MCP server or agent plugin before allowing it?
DISCUSSION · 0

No comments yet. Share a correction or a different way to answer it.