A Kubernetes cluster runs a mix of CPU and GPU workloads. A GPU training job is stuck in 'Pending' state. You check the node status and see that several GPU nodes have available GPU capacity. However, the pod has a nodeSelector for a specific GPU model that is not present on any node. What is the most likely cause?

The pod has a toleration for a taint that is not present.

The GPU resource name is misspelled.

The nodeSelector does not match any node's labels.

The pod's resource requests exceed the available GPU memory.

Which Kubernetes feature allows you to increase the number of replicas of an AI model-serving deployment to handle more inference requests?

Creating a new Namespace for each request

Horizontal scaling by updating the replicas field in the Deployment

Modifying the Service type to NodePort

Vertical scaling by increasing the pod's CPU limits

A data science team runs a distributed training job on a Kubernetes cluster. Each worker pod needs a dedicated 200 GiB volume to store checkpoints. The cluster admin has configured a Trident backend named 'ontap-nfs' and created a storage class 'fast-nfs'. The team wants to ensure that each worker pod automatically gets its own volume when the pod is created, without manual volume creation. What should the team do?

Define an emptyDir volume in the pod spec for each worker.

Manually create a PersistentVolume (PV) for each worker pod and reference it in the pod spec.

Use a StatefulSet with a volumeClaimTemplate referencing the 'fast-nfs' storage class.

Create a single PersistentVolumeClaim (PVC) and share it among all worker pods.

NetAppCertified AI Expert

Domain 3Objective 4

Demonstrate the Ability to Execute AI Workloads at Scale with Kubernetes (Trident) CERTIFIED-AI-EXPERT Practice Questions (Page 4)

Part of the AI Software Architectures domain, which accounts for 18% of the CERTIFIED-AI-EXPERT exam. NetApp does not publish an official question count, but from its 90-minute exam (~35–60 total, ~6–11 in this domain), expect 1–2 from this objective — we provide 36 practice questions to prepare you well beyond it. (estimate)

36questions here

8free pages

9concepts

18%of the exam

Questions 16–20

16expert · hard
A Kubernetes cluster runs a mix of CPU and GPU workloads. A GPU training job is stuck in 'Pending' state. You check the node status and see that several GPU nodes have available GPU capacity. However, the pod has a nodeSelector for a specific GPU model that is not present on any node. What is the most likely cause?
Select an answer first
17foundation · easy
Which Kubernetes feature allows you to increase the number of replicas of an AI model-serving deployment to handle more inference requests?
Select an answer first
18foundation · easy
Which Trident feature allows you to create point-in-time, space-efficient copies of a persistent volume to protect AI training data?
Select an answer first
19application · medium
A data science team runs a distributed training job on a Kubernetes cluster. Each worker pod needs a dedicated 200 GiB volume to store checkpoints. The cluster admin has configured a Trident backend named 'ontap-nfs' and created a storage class 'fast-nfs'. The team wants to ensure that each worker pod automatically gets its own volume when the pod is created, without manual volume creation. What should the team do?
Select an answer first
20foundation · easy
Which field in a PersistentVolumeClaim is used to request a specific StorageClass for dynamic provisioning with Trident?
Select an answer first

Finished these 5 questions?

Review the revealed explanations, or continue through the curriculum.

Next 5 questions →Back to CERTIFIED-AI-EXPERT

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-AI-EXPERT” is a trademark of its owner, used for identification only.

Demonstrate the Ability to Execute AI Workloads at Scale with Kubernetes (Trident) CERTIFIED-AI-EXPERT Practice Questions (Page 4)

Questions 16–20

Which Kubernetes feature allows you to increase the number of replicas of an AI model-serving deployment to handle more inference requests?

Which Trident feature allows you to create point-in-time, space-efficient copies of a persistent volume to protect AI training data?

Which field in a PersistentVolumeClaim is used to request a specific StorageClass for dynamic provisioning with Trident?

Keep studying — AI Software Architectures