You are on call. At 09:14 an alert fires: the pricing model's predictions have shifted sharply. The mean predicted price has jumped 22% in the last hour. Nothing has been deployed for six days.
Downstream, the pricing service is using these predictions live, and the finance team has already noticed anomalous quotes. Latency and error rates are normal. Prediction volume is normal.
You have: prediction logs with feature vectors, the feature store, upstream data pipeline dashboards, model registry, and the ability to roll back or disable the model in favour of a rules-based fallback.
Walk me through the first 30 minutes. Be specific about what you check, in what order, and what decision you make at each point.
Build an interview-ready technical plan: frame the problem, explain the ordered approach, name the trade-offs, and show how you would validate the outcome. The AI reviewer grades that reasoning against this problem’s rubric.
Four guided sections · works on any screen