Practice for the Red Hat Certified Specialist in OpenShift AI (EX267) exam with 18 exam-style practice questions, instant answer reveals, and concise explanations of every correct answer. Topics include: What type of OpenShift resource is automatically created when you add a Data Connection to an OpenShift AI project?. Follow @CertPunch and visit certpunch.com for more certification practice exams and study content.
Prefer hands-on? Take this round as an interactive practice test — answer every question, get instant feedback, and see your score: Start the Red Hat Certified Specialist in OpenShift AI (EX267) practice test →
What you will practice
- What type of OpenShift resource is automatically created when you add a Data Connection to an OpenShift AI pr…
- What is the function of the ModelMesh routing layer when an external client sends a request to a multi-model…
- What cluster-level role must a user possess to install the Red Hat OpenShift AI Operator from the OperatorHub?
- What happens if the underlying S3 credentials in a Data Connection are updated after a model server has alrea…
- What is the primary purpose of a Data Connection in Red Hat OpenShift AI?
- To build and import a custom notebook image into OpenShift AI, where must the container image be stored so th…
Answers and explanations
Tap a question to expand the answer and the exam reasoning. Try to commit to your own pick first.
Q1. What type of OpenShift resource is automatically created when you add a Data Connection to an OpenShift AI project?
Answer: D. Kubernetes Secret
Data Connections are stored as Kubernetes Secrets so that workbenches and model servers can securely mount S3 credentials. ConfigMaps are not used because they lack the encryption needed for sensitive AWS keys and endpoints.
Q2. What is the function of the ModelMesh routing layer when an external client sends a request to a multi-model endpoint?
Answer: A. It intelligently routes the inference request to the specific container instance where the target model is cached or loaded
The routing layer intercepts external requests and forwards them to the correct pod where the requested model resides. It manages model caching and lifecycle, rather than compressing payloads or handling browser redirects.
Q3. What cluster-level role must a user possess to install the Red Hat OpenShift AI Operator from the OperatorHub?
Answer: B. cluster-admin
Installing operators from OperatorHub requires cluster-admin privileges to create the necessary namespace and operator groups. Standard project admin rights only allow deployment of resources within an existing namespace.
Q4. What happens if the underlying S3 credentials in a Data Connection are updated after a model server has already successfully deployed a model?
Answer: C. The running model continues to serve, but subsequent re-deployments or updates will fail
Running models are already loaded into memory and do not constantly re-read the S3 credentials. However, new pods launched during updates or scale events will fail to pull the model if the new credentials are invalid.
Q5. What is the primary purpose of a Data Connection in Red Hat OpenShift AI?
Answer: B. To provide easy and secure access to cloud object storage resources
Data Connections provide a streamlined way to securely pass S3-compatible object storage credentials to notebooks and model servers. They abstract the manual creation of environment variables or secrets for the end user.
Q6. To build and import a custom notebook image into OpenShift AI, where must the container image be stored so the cluster can pull it?
Answer: D. An accessible container registry like Quay.io
Custom notebook images must be pushed to an OpenShift-accessible container registry like Quay.io so the cluster can pull them. OpenShift cannot natively build or pull container images directly from local directories or S3 buckets.
Q7. What is the maximum number of models that can be served reliably on a single model server instance when utilizing KServe Single-Model Serving?
Answer: B. Exactly 1 model per server instance
KServe Single-Model Serving deploys exactly one model per server instance to isolate resources completely. This differs from ModelMesh, which uses high-density packing to serve multiple models within a shared deployment.
Q8. What is the main advantage of utilizing ModelMesh for multi-model serving?
Answer: A. High-density packing of multiple small models on minimal compute resources
ModelMesh intelligently packs multiple models into shared runtime pods to maximize memory and compute efficiency. This makes it ideal for managing many small models, whereas KServe dedicates isolated resources per deployment.
Q9. In OpenShift AI, a user creates an environment for notebook execution. What is this environment called?
Answer: C. Workbench
A workbench is the designated environment in OpenShift AI for developing and executing notebook code. While a Data Science Project houses the resources, the workbench is the specific pod running the Jupyter environment.
Q10. What underlying technology is leveraged by OpenShift AI's Multi-Model serving to optimize memory by packing multiple models onto few pods?
Answer: B. ModelMesh
ModelMesh is the underlying technology for multi-model serving that efficiently packs multiple models into fewer pods to optimize memory. KServe is typically used for single-model serving or large models needing dedicated resources.
Q11. What does a status of Ready indicate on an OpenShift AI InferenceService custom resource?
Answer: A. The model has successfully loaded into memory and is ready to receive network inference traffic
A Ready status on an InferenceService confirms the model is fully loaded into memory and prepared to handle incoming network inference traffic. Training happens within workbenches or pipelines, not during inference.
Q12. Which dashboard menu section allows a cluster administrator to view and create new Accelerator Profiles for user access?
Answer: B. Settings -> Accelerator profiles
Cluster administrators manage and create Accelerator Profiles under the Settings section of the dashboard. These profiles map hardware accelerators like GPUs to user workbenches for specialized compute tasks.
Q13. What is the primary underlying Kubernetes operator used to deploy Red Hat OpenShift AI?
Answer: D. Red Hat OpenShift AI Operator
The Red Hat OpenShift AI Operator is the primary operator responsible for deploying and managing the platform components. The older OpenShift Data Science Operator name is deprecated.
Q14. What happens to a user's unsaved code inside an OpenShift AI Jupyter notebook if the notebook pod is restarted or deleted, assuming a PVC is mounted?
Answer: D. Code saved within the persistent volume directory (/opt/app-root/src) is preserved
Any files explicitly saved to the persistent volume mount directory survive pod restarts or deletions. Unsaved data in memory is lost, but work inside the PVC path is preserved.
Q15. What configuration item specifies the access keys needed to download model weights from an isolated private storage repository?
Answer: A. An OpenShift Secret referenced by the project's Data Connection configuration
Data Connections map to OpenShift Secrets containing endpoint and credential details for private storage. Placing credentials in a public ConfigMap is insecure, and private SSH keys are not the mechanism used for S3-compatible data access.
Q16. What must a cluster administrator configure to allow OpenShift AI to utilize NVIDIA GPUs?
Answer: C. NVIDIA GPU Operator
The NVIDIA GPU Operator handles provisioning the necessary drivers and device plugins for OpenShift AI workloads. Node Tuning is for performance tweaks, and SRIOV is for network hardware.
Q17. How can you share a Data Science Project with another user in your organization via the OpenShift AI dashboard?
Answer: A. Using the "Permissions" tab to grant access to users or groups
The dashboard uses a Permissions tab to manage role bindings for sharing projects safely. Exporting Helm charts or copying admin tokens bypasses the intended UI integration and violates standard security practices.
Q18. Which tool is used to compile a Python-based pipeline defined with the Kubeflow Pipelines SDK into a YAML format file for OpenShift AI?
Answer: A. kfp.compiler.Compiler()
The kfp.compiler.Compiler class compiles Python-based pipelines into a YAML file. The kfp.client.Client class is used to connect to the pipeline API server and submit runs.
More Red Hat Certified Specialist in OpenShift AI (EX267) drills and other practice exams are on @CertPunch. New rounds drop every few days at certpunch.com.