Section 20 of 24 · 5 lessons
Deploying and Maintaining Models
A model needs to keep behaving as intended after it leaves a training script. Botie will save the full prediction procedure, check incoming records, preserve model versions, and inspect later performance. We will also work through what a useful rollback must restore.
From Training to a Repeatable Prediction
Saved weights need the preparation and feature meanings that made them useful. Preserve that procedure and validate the records entering it.
Running and Identifying a Model
A prediction should stay consistent across requests and traceable to a particular version. Test batch behaviour and preserve complete artifacts before changing the served model.
Maintaining the Prediction Service
New inputs and later outcomes can reveal different problems. Monitor data, accuracy, and service behaviour separately, with a tested response when something changes.