Practice free →
HomeGCP ML Engineercloudcomputingmleserving › Which GCP-compatible streaming runtime should an…

Which GCP-compatible streaming runtime should an ML Engineer use to MAKE PREDICTIONS over a Pub/Sub stream in real-time — applying a trained model to each event?

ACloud Memorystore
BManual polling of Pub/Sub by a developer
CCloud Dataflow (with a Beam pipeline that calls a Vertex AI endpoint or runs an in-pipeline model)
DCloud DNS
Answer & Solution
Correct answer: C. Cloud Dataflow (with a Beam pipeline that calls a Vertex AI endpoint or runs an in-pipeline model)
Dataflow + Vertex AI is the streaming-inference pattern (per PMLE §4.1). The others aren't streaming runtimes.
Solve this in the app — GCP ML Engineer practice & 24k+ MCQs →
Related questions