Question

Your team runs a fraud-detection model on a Vertex AI endpoint with model monitoring enabled for feature drift. You want production drift alerts to automatically kick off your existing Vertex AI Pipelines retraining workflow, without a human manually starting the run. You also want to avoid standing up and maintaining any always-on infrastructure. Which approach best achieves this automated, event-driven retraining trigger?

Accepted Answer

Configure the monitoring job to publish drift alerts to a Pub/Sub topic, subscribe a Cloud Function to that topic, and have the function call the Vertex AI Pipelines API to submit the retraining pipeline run.

Answer

Create a Cloud Scheduler cron job that runs the retraining pipeline every hour so it always retrains shortly after any drift occurs.

Answer

Deploy a long-running Compute Engine VM that continuously polls the monitoring job's metrics and launches the pipeline when drift thresholds are exceeded.

Answer

Enable pipeline component caching so that when drift is detected the pipeline automatically re-executes only the changed steps.