A media company is training a video analysis model using a large dataset of high-resolution videos. The training cluster uses GPUs with NVMe local storage. The data is currently stored on a central NAS. The team notices that GPU utilization is low because data loading is slow. They want to improve performance without adding more GPUs. Which solution is most effective?

Increase the number of CPU cores in each server

Move the dataset to object storage

Use a slower but larger storage system

Implement GPU direct storage to allow GPUs to access data directly from NVMe storage

Which governance practice helps ensure that an AI model's predictions comply with regulatory requirements?

Using the largest possible training dataset

Deploying the model on-premises only

Regularly auditing model decisions and documenting the model's behavior

Increasing the model's inference speed

A university research team is setting up a new AI cluster for training large neural networks. They need to choose hardware that will minimize training time. Which combination of infrastructure components is most critical for this workload?

Large memory in the CPU and a high-resolution monitor

High-capacity hard drives, standard Ethernet, and a single powerful CPU

High-performance GPUs, high-throughput storage, and low-latency networking

Multiple GPUs but with low-bandwidth network connections

A healthcare provider is using an AI model to predict patient readmission risk. The model was trained on patient data that includes protected health information (PHI). The provider wants to deploy the model in a public cloud to reduce costs, but must comply with HIPAA. Which action is required?

Sign a Business Associate Agreement (BAA) with the cloud provider and ensure the cloud environment is configured for HIPAA compliance

Anonymize the data before training so that HIPAA no longer applies

Use a private cloud within the provider's data center to avoid HIPAA

Deploy the model without any additional agreements because the cloud provider is responsible for compliance

A startup is developing a natural language processing model that requires 100 GPUs for training. They have a limited budget and cannot afford to purchase the hardware. They also need to keep the training data within the country for regulatory reasons. Which option is most viable?

Use a hybrid approach: train on-premises with 10 GPUs and use cloud for the rest

Reduce the model size to fit on a single GPU

Use a public cloud provider that offers GPU instances in a region within the country

Purchase 100 GPUs and build an on-premises cluster

NetAppCertified Technology Solutions Professional (NCTSP)

Domain 1Objective 4

AI Concepts CERTIFIED-TECHNOLOGY-SOLUTIONS-PROFESSIONAL Practice Questions (Page 4)

Part of the Infrastructure domain, which accounts for 25% of the CERTIFIED-TECHNOLOGY-SOLUTIONS-PROFESSIONAL exam.

34questions here

7free pages

9concepts

25%of the exam

Questions 16–20

16expert · hard
A media company is training a video analysis model using a large dataset of high-resolution videos. The training cluster uses GPUs with NVMe local storage. The data is currently stored on a central NAS. The team notices that GPU utilization is low because data loading is slow. They want to improve performance without adding more GPUs. Which solution is most effective?
Select an answer first
17foundation · easy
Which governance practice helps ensure that an AI model's predictions comply with regulatory requirements?
Select an answer first
18application · medium
A university research team is setting up a new AI cluster for training large neural networks. They need to choose hardware that will minimize training time. Which combination of infrastructure components is most critical for this workload?
Select an answer first
19expert · hard
A healthcare provider is using an AI model to predict patient readmission risk. The model was trained on patient data that includes protected health information (PHI). The provider wants to deploy the model in a public cloud to reduce costs, but must comply with HIPAA. Which action is required?
Select an answer first
20expert · hard
A startup is developing a natural language processing model that requires 100 GPUs for training. They have a limited budget and cannot afford to purchase the hardware. They also need to keep the training data within the country for regulatory reasons. Which option is most viable?
Select an answer first

Finished these 5 questions?

Review the revealed explanations, or continue through the curriculum.

Next 5 questions →Back to CERTIFIED-TECHNOLOGY-SOLUTIONS-PROFESSIONAL

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 NetApp. “CERTIFIED-TECHNOLOGY-SOLUTIONS-PROFESSIONAL” is a trademark of its owner, used for identification only.