Your question is Debugging Distributed Failures. 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).
Describe your approach to debugging a system failure in a distributed backend environment.
Explain how you would scope the incident, use logs, metrics, traces, and dependency health signals to isolate the fault, and distinguish a root cause from downstream symptoms. Discuss immediate mitigation, validation, communication, and follow-up actions, including how you would detect feature drift, training-serving skew, or model-serving failures when the backend includes ML components.