Lab: September replayed through a monitoring job
Module 11 · Lab · About 60 minutes · Needs: lessons 1 and 2 of this module, and the Module 10 lab (or nothing: Step 1 rebuilds Modules 1 to 10)
What you are building
Nothing crashed in September, and nobody knew the model had gone wrong. This lab builds the job that would have known: pipelines/monitor_flow.py reads every scored parcel, computes the lessons' health numbers for each ship day and each morning's matured labels, and turns them into pages, tickets and dashboard lines with thresholds tuned on history. Then it writes the three documents that make monitoring a practice: the dashboard specification, the postmortem for 7 September, and ADR-008.
The lessons monitored a stand-in model trained in the notebook. The lab monitors the real one: model v2 from artifacts/model.onnx, served the way the service ran in September, before contract 1.1's grams fix reached it. Every number below is v2's, replayed from 6 July to 27 September. No cell calls a hosted model.
Open a new Colab notebook, call it module-11-lab.ipynb, and run each cell in order.