Skip to content
AI Security

Interview question · Cloud, Container & AI Infrastructure Security

Design backup, rollback and recovery for an AI system's model, prompts, retrieval index and tool configuration.

An AI system's behaviour lives in more places than its code. The scored answer versions every one of them together, so you can return to a known-good state in minutes after a bad change or a poisoning.

mediumNewPreparation tags: Amazon, Microsoft, GoogleUpdated Nov 2026

An AI system's behaviour lives in more places than its code. The scored answer versions every one of them together, so you can return to a known-good state in minutes after a bad change or a poisoning.

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

A bad model update and a corrupted vector index took an AI assistant down for a day, and nobody could say which combination had last worked.

  1. What determines an AI feature's behaviour?
  2. How should releases be recorded?
  3. What should be backed up, and where?
  4. What if the index was poisoned rather than corrupted?
  5. How do you know rollback works?
  6. What recovery design would you implement?

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 trackHow do infrastructure as code and policy as code prevent misconfiguration?Next in this trackHow do you use cloud audit logs to detect credential misuse and data exfiltration?Next in this trackHow do you run vulnerability and patch management for a GPU fleet used for training and inference?
DISCUSSION · 0

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