Your question is High Availability and DR Design. Take a moment with it on the right.
Talk me through your thinking if you like. When you're confident, submit your answer and I'll grade it like a real screen (7/10 or better passes).
How do you design systems for high availability and disaster recovery? Explain how you would design an end-to-end ML platform covering online inference, batch pipelines, model artifacts, feature data, regional failover, backups, recovery objectives, and operational testing. Address the tradeoffs between availability, consistency, recovery time, recovery point, cost, and model quality.