AI Governance / Post-Deployment Monitoring

Model Lifecycle Monitor

Deploying an AI model is not the end of the product lifecycle. AI systems can change in behavior over time as user interactions, data distributions, and operating environments evolve. This project demonstrates a continuous governance approach by monitoring simulated model drift against predefined thresholds, automatically flagging deviations, documenting governance actions, and reverting the model to a previously approved state when necessary.

W00

Lifecycle State

Week
Model version
Overall PSI
Current state

Per-Segment Drift (PSI by Week)

Threshold bands

Stable
PSI < 0.10
Drift watch
0.10 – 0.25
Breach
PSI ≥ 0.25
Hover or focus a cell for its segment, week, and PSI value. Where drift occurs, it localizes to specific behavior segments.

Audit Trail

Overall PSI across 12 weeks

Governance log

Run the simulation to populate the audit trail.