PMLE Automating and Orchestrating ML Pipelines Practice Question

Answer analysis

Option-by-option breakdown

For each option: why learners choose it and why it is or isn't the right answer here.

• ✓

  Use Cloud Storage Pub/Sub notifications to send events to a Cloud Function that triggers the pipeline.

  Why this is correct

  Cloud Storage Pub/Sub notifications emit an event whenever an object lands in the bucket, and a Cloud Function subscribed to that topic invokes the Vertex AI Pipeline. This satisfies the stem's requirement to run only when new training data actually arrives, avoiding polling.

• ✗

  Use Cloud Tasks to queue a pipeline run whenever a new file is uploaded.

  Why it's wrong here

  Cloud Tasks dispatches HTTP requests to a target service; it has no native Cloud Storage trigger and cannot itself detect uploads, so nothing enqueues a run. It suits rate-limited or delayed delivery of explicit tasks. A Cloud Storage object-finalise event notification is what detects the new file.

• ✗

  Configure the pipeline to run on a schedule and check for new data inside the pipeline.

  Why it's wrong here

  A schedule fires the pipeline regardless of whether data exists, so runs occur with no new training data and the check inside the pipeline only wastes execution time. Scheduled pipelines suit periodic retraining on stable datasets. Object-finalise events from the bucket are what signal genuine data arrival.

• ✗

  Set up a Cloud Scheduler job that runs every minute to check for new files.

  Why it's wrong here

  Polling every minute adds up to sixty seconds of latency and invokes checks when nothing has arrived, so the pipeline is not truly event-driven. Cloud Scheduler suits fixed-time batch jobs, such as nightly retraining. A Cloud Storage object-finalise event routed to the pipeline is the correct trigger.

Go deeper

Related to this question