Skip to content
AI Security

Interview question · Cloud, Container & AI Infrastructure Security

How would you control egress and detect exfiltration from a training cluster?

Training clusters hold the data and the weights, and run code from hundreds of dependencies. Default-deny egress is the control that holds when everything else is fooled.

hardNewPreparation tags: Anthropic, OpenAI, Google DeepMindUpdated Nov 2026

Training clusters hold the data and the weights, and run code from hundreds of dependencies. Default-deny egress is the control that holds when everything else is fooled.

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.

signing in unlocks more free answers per topic

What if · the follow-ups in this answer

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

A training cluster holds proprietary data and model weights, and jobs can currently reach any internet address.

  1. What should the default be?
  2. How do you handle the rest?
  3. What if researchers need internet access for datasets?
  4. How do you detect exfiltration attempts?
  5. What remains after these controls?
  6. What egress design would you implement?

Each rung has a nudge, a model answer, the principle that carries over and the weak answer to avoid. Sign in free to open it.

UP NEXT ON YOUR JOURNEY
Next in this trackWhy are exposed notebooks and distributed compute dashboards dangerous, and how do you prevent it?Next in this trackHarden a Kubernetes cluster that runs GPU training jobs.Next in this trackWhat isolation and side-channel concerns arise when GPUs are shared between tenants?
DISCUSSION · 0

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