HikeLetter | PMP Training & Certification Online Course | Project Management Learning

1

An AI platform operations team is reviewing a production cluster and asks: What is the main purpose of GPU monitoring in an AI infrastructure environment?

A. Track utilization, memory, temperature, errors, and performance B. Replace access control C. Guarantee model accuracy D. Eliminate capacity planning

2

An AI platform operations team is reviewing a production cluster and asks: Which metric best represents the amount of GPU memory currently in use?

A. GPU memory utilization or allocated memory B. DNS query count C. Number of user passwords D. HTTP status text length

3

An AI platform operations team is reviewing a production cluster and asks: Why is a container image used in AI infrastructure?

A. Package an application and its dependencies consistently B. Guarantee zero vulnerabilities C. Replace all host operating systems D. Provide unlimited GPU memory

4

An AI platform operations team is reviewing a production cluster and asks: What is the role of a container orchestrator such as Kubernetes?

A. Schedule and manage containerized workloads across a cluster B. Label training data automatically in every case C. Replace network routing entirely D. Guarantee that every pod has a GPU

5

An AI platform operations team is reviewing a production cluster and asks: Why use a GPU operator or device plugin in a cluster?

A. Expose and manage GPU resources for workloads B. Replace all security policies C. Create training labels D. Convert every workload to CPU-only

6

An AI platform operations team is reviewing a production cluster and asks: What is a common benefit of Infrastructure as Code (IaC)?

A. Repeatable, reviewable provisioning from version-controlled definitions B. Untracked manual changes C. Guaranteed zero outages D. Elimination of monitoring

7

An AI platform operations team is reviewing a production cluster and asks: What is a job scheduler responsible for in an AI cluster?

A. Allocate resources and schedule jobs according to policies and availability B. Write model prompts automatically C. Encrypt every dataset without configuration D. Replace storage systems

8

An AI platform operations team is reviewing a production cluster and asks: A GPU workload is slower than expected. What should be investigated first?

A. Metrics, utilization, memory, logs, and workload bottlenecks B. Delete the cluster immediately C. Disable monitoring D. Upgrade every node without measurement

9

An AI platform operations team is reviewing a production cluster and asks: What is the purpose of a high-speed interconnect in multi-GPU training?

A. Reduce communication overhead between accelerators or nodes B. Increase DNS cache lifetime C. Replace model evaluation D. Store user credentials

10

An AI platform operations team is reviewing a production cluster and asks: Why is checkpointing useful during long-running model training?

A. Resume from saved state after interruption B. Guarantee no training errors C. Remove the need for storage D. Increase network packet loss

AI Infrastructure & Operations : NCA-AIIO

Overview

Course Description

Course Questions