What is the purpose of VPC Service Controls when applied to a model serving endpoint?

To prevent data exfiltration by restricting access to the endpoint from outside the VPC

To enable load balancing across multiple regions

To allow the endpoint to scale automatically

To increase the throughput of the endpoint

A company has separate teams for training and serving. The training team computes features during training, and the serving team recomputes the same features at inference time. This has led to training-serving skew. They want to centralize feature management. What should they do?

Use a feature store only for training and keep serving features separate

Store all features in Vertex AI Feature Store and use the same feature values for both training and serving

Have the serving team use the training team's feature computation code

Store features in a shared BigQuery table for both training and serving

Which benefit does a feature store provide when multiple teams develop models using overlapping data?

It enables feature reuse and consistency across different models and teams

It guarantees lower inference latency for all models

It automatically retrains models when new data arrives

It eliminates the need for data validation and monitoring

What is the purpose of traffic splitting when deploying a new model version to an endpoint?

To route a percentage of requests to the new version for testing while keeping the rest on the stable version

To increase the total throughput of the endpoint

To reduce the latency of the endpoint by distributing load

To automatically retrain the model based on incoming requests

What is a key cost trade-off when choosing between CPU and GPU instances for serving?

CPUs have higher per-hour cost but lower latency

GPUs have higher per-hour cost but can provide lower latency for compute-intensive models

GPUs are always cheaper than CPUs

Google CloudProfessional Machine Learning Engineer

Domain 4Objective 2

Scaling Online Model Serving PROFESSIONAL-MACHINE-LEARNING-ENGINEER Practice Questions (Page 4)

Part of the Serving and scaling models domain, which accounts for ~20% of the PROFESSIONAL-MACHINE-LEARNING-ENGINEER exam.

37questions here

8free pages

14concepts

~20%of the exam

14 Concepts in This Objective

Feature Store fundamentals
Serving features from Feature Store
Feature Store configuration and monitoring
Public and private endpoints
Deploying models to endpoints
Endpoint security and access control
Hardware selection for serving
Hardware optimization and cost trade-offs
Scaling serving infrastructure
Containerized serving and inference platforms
Throughput and latency management
Model tuning for production
Serving-specific model optimization
Production model lifecycle management

Questions 16–20

16foundation · easy
What is the purpose of VPC Service Controls when applied to a model serving endpoint?
Select an answer first
17application · medium
A company has separate teams for training and serving. The training team computes features during training, and the serving team recomputes the same features at inference time. This has led to training-serving skew. They want to centralize feature management. What should they do?
Select an answer first
18foundation · easy
Which benefit does a feature store provide when multiple teams develop models using overlapping data?
Select an answer first
19foundation · easy
What is the purpose of traffic splitting when deploying a new model version to an endpoint?
Select an answer first
20foundation · easy
What is a key cost trade-off when choosing between CPU and GPU instances for serving?
Select an answer first

Finished these 5 questions?

Review the revealed explanations, or continue through the curriculum.

Next 5 questions →Back to PROFESSIONAL-MACHINE-LEARNING-ENGINEER

Free Basic Practice is a study aid with revealable answers — not a scored exam. Examers.io is independent and not affiliated with or endorsed by Google Cloud. “PROFESSIONAL-MACHINE-LEARNING-ENGINEER” is a trademark of its owner, used for identification only.