Your question is Handle Failures and Retries. 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 would you handle failures and retries?
Asked in the R4 Hiring Manager stage.
Focus on what should be retried, when to fail open or fail closed, and how you would avoid duplicate side effects or cascading retries. Keep the answer practical and specific to a production ML serving workflow.