Your question is Real-Time ML Model Monitoring. 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 design a system to monitor the performance of a machine learning model in real-time?
Explain how you would monitor serving health, input and prediction distributions, data quality, feature drift, training-serving skew, and model quality when labels may arrive later. Cover the online architecture, alerting strategy, dashboards, feedback loops, rollout safeguards, and responses to detected regressions. State the clarifying questions you would ask about traffic, latency, model type, label delay, and reliability requirements.