Essential session storage is required to sign in. Optional analytics remain off unless you allow them.
Serve a trained model behind a FastAPI endpoint: schema-validated input/output, train/serve-shared preprocessing, version pinned from a registry, every prediction logged, input-drift and prediction-drift checks against the training snapshot, and a shadow-mode comparison script for the next candidate version.
Preprocessing shared between training and serving; version pinned and reported per prediction.
Drift checks on inputs and scores with thresholds; prediction log designed for label backfill.
Validated I/O, health/readiness, warm-up, containerized.
Shadow comparison exists and was run; rollback is a config change.
Use this README structure to make the work reviewable.
DevPath fetches a read-only, size-capped snapshot. Submitted code is never executed.
# Deploy and monitor a model as a service ## Goal Serve a trained model behind a FastAPI endpoint: schema-validated input/output, train/serve-shared preprocessing, version pinned from a registry, every prediction logged, input-drift and prediction-drift checks against the training snapshot, and a shadow-mode comparison script for the next candidate version. ## Acceptance criteria - [ ] Train/serve parity: Preprocessing shared between training and serving; version pinned and reported per prediction. - [ ] Monitoring: Drift checks on inputs and scores with thresholds; prediction log designed for label backfill. - [ ] Service quality: Validated I/O, health/readiness, warm-up, containerized. - [ ] Rollout discipline: Shadow comparison exists and was run; rollback is a config change. ## Verification - [ ] Document installation and run commands. - [ ] Record automated checks and their exact commands. - [ ] Add representative output, screenshots, or a short demo where useful. - [ ] Confirm failure paths and known constraints. ## Decision log ### Decision title - Context: - Choice: - Alternatives considered: - Tradeoffs: ## Evidence - Link each rubric criterion to the file, test, or artifact that demonstrates it. ## Known limitations - List what is intentionally out of scope and what should be improved next.