Objective 4.1

PMLE

Serving models

Objective 4.1 sits in Serving and scaling models, which carries 20% of the ML Engineer exam. The questions below are original, written from the official objective title above, and each explanation cites the Google Cloud page it rests on.

Objective title verbatim from the official objectives. Google Cloud exam page

A worked example

Shown solved, with the whole explanation open: this is what every question here carries.

4-1Serving and scaling models

A team sets both a starting and a maximum replica count on a batch inference job and sees the ceiling ignored. What governs the replica count?

The maximum, once the queue depth growsThere is no queue to grow during the run.
The number of input files suppliedFile count matters for partitioning, not for scaling.
The machine type chosen for the jobMachine type sets speed per replica, not their number.
The starting count, with no autoscalingCorrect · your answerCorrect. The split happens once at launch.

Correct.

Concept

Elastic scaling exists to react to demand nobody knew in advance. When the whole workload is present before the first replica starts, the system can partition it once and never needs to react at all.

Why D

Unlike online inference, batch inference jobs do not autoscale. All of the input data is known up front, so the system partitions the data to each replica when the job starts using the starting_replica_count parameter, and max_replica_count is ignored.

Source

Unlike online inference, batch inference jobs don't autoscale . Because all of the input data is known up front, the system partitions the data to each replica when the job starts. The system uses the starting_replica_count parameter. The max_replica_count parameter is ignored.

Get batch inferences from a custom trained model, Agent Platform documentation, checked August 2026
#gcp#batch-inference#scaling#serving

Now you: objective 4.1 questions

No account needed. The explanation opens when you answer.

Sample question 1 of 3

4-1Serving and scaling models

A team stores batch input in a bucket in europe-west4 while the model itself sits in us-central1. What does the placement rule require?

Sample question 2 of 3

4-1Serving and scaling models

A throughput sensitive batch job reads CSV input and its replicas sit idle waiting for work. Which input format would partition automatically?

Sample question 3 of 3

4-1Serving and scaling models

A team grants a custom service account to a batch job and finds it is not used when the source table is read. Which identity reads it?

Full ML Engineer question bank coming

We’re writing the complete bank from the official objectives right now. Leave your email and we’ll tell you when it ships, nothing else, ever.

Read the sources

These are the official pages the questions above cite. Reading them is studying the objective from the primary source, which is what the explanations point you toward anyway.

More objectives in Serving and scaling models