Your question is Monitoring for Drift and Data Quality. 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 structure a monitoring pipeline to track model drift and data quality in production?
Explain the signals you would collect, how you would compare production data with training and serving baselines, and how alerts would lead to investigation or automated mitigation. Cover both cases where ground-truth labels are delayed or unavailable and how you would avoid alert fatigue.