Your question is Failure-Resilient Microservices. 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 for failure in a microservices architecture?
Treat this as a practical ML platform design exercise. Explain how an online inference path, asynchronous pipelines, dependencies, and model-serving components continue operating during partial failures, and how you detect, contain, and recover from them.
