Practice for the Red Hat Certified Specialist in OpenShift AI (EX267) exam with 17 exam-style practice questions, instant answer reveals, and concise explanations of every correct answer. Topics include: What type of OpenShift resource must be configured to allow external traffic to reach a model served via KServe?. 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 must be configured to allow external traffic to reach a model served via KSer…
- Which runtime provides the traditional inference execution base engines for serving scalable open-source ONNX…
- Which API object type specifies the compute resource limits enforced for an automated data science workbench…
- Which core service handles inference sourcing from multiple independent compute engines simultaneously within…
- WHAT REPRESENTS THE PRINCIPAL WAY TO UPDATE THE REPOSITORY SOURCE FILE DEFINITIONS INSIDE A WORKING JUPYTERLA…
- WHAT REPRESENTS THE HIGHEST LAYER CONTROL MECHANISM USED BY CLUSTER ADMINS TO RESTRICT THE SIZE OF DATA SCIEN…
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 must be configured to allow external traffic to reach a model served via KServe?
Answer: B. Route or Gateway
Serverless KServe Inference Services typically expose external traffic through an OpenShift Route or an Istio Gateway. Be careful with Ingress, as OpenShift serverless prioritizes Knative routing and Routes for direct external access.
Q2. Which runtime provides the traditional inference execution base engines for serving scalable open-source ONNX specification models?
Answer: A. OpenVINO Model Server runtime container images
OpenVINO Model Server is the underlying runtime used in OpenShift AI to serve ONNX models efficiently. While other frameworks like PyTorch can export to ONNX, OpenVINO is the specific target engine designed for optimized, scalable inference of that format.
Q3. Which API object type specifies the compute resource limits enforced for an automated data science workbench environment?
Answer: B. Notebook CRD specification resources map
The Notebook custom resource definition contains the specific resource requests and limits map that dictate the compute allocation for the pod. LimitRanges apply cluster-wide defaults, but the Notebook CRD directly targets the workbench environment.
Q4. Which core service handles inference sourcing from multiple independent compute engines simultaneously within a ModelMesh deployment?
Answer: C. ModelMesh Orchestrator routing container wrapper
The ModelMesh Orchestrator acts as the central routing wrapper that manages lifecycle and inference requests across various built-in runtime engines. Kube-Scheduler only handles initial pod placement to nodes, not dynamic inference traffic distribution.
Q5. WHAT REPRESENTS THE PRINCIPAL WAY TO UPDATE THE REPOSITORY SOURCE FILE DEFINITIONS INSIDE A WORKING JUPYTERLAB ENVIRONMENT?
Answer: A. Git integration clone and pull routines
Using Git integration to clone and pull updates is the principal way to synchronize repository source files inside a JupyterLab Workbench. Modifying persistent volumes directly bypasses version control and risks severe data corruption.
Q6. WHAT REPRESENTS THE HIGHEST LAYER CONTROL MECHANISM USED BY CLUSTER ADMINS TO RESTRICT THE SIZE OF DATA SCIENCE PERSISTENT VOLUMES?
Answer: C. ResourceQuota on the project namespace
Applying a ResourceQuota to the project namespace is the highest layer control mechanism to restrict persistent volume sizes. LimitRange objects set constraints on individual compute pods, not the aggregate storage consumption.
Q7. Which UI module facilitates comparing historical pipeline run accuracy metrics side-by-side within OpenShift AI?
Answer: B. Pipeline Runs Details Metrics and Runs Comparison interface dashboard panels
The Pipeline Runs Details interface provides specific dashboard panels to compare historical run metrics side-by-side. Grafana primarily visualizes cluster infrastructure health rather than individual model training accuracy metrics.
Q8. Which engine manages secure token authentication and authorization traffic control for KServe served ML models?
Answer: D. Authorino
Authorino is the correct choice because OpenShift AI uses it to enforce token-based authentication and authorization policies for KServe inference services. While OAuth Proxy handles generic OpenShift routing security, Authorino provides the specialized external API authentication.
Q9. Which S3 compliant bucket property must be declared to connect an on-premises Ceph endpoint as a data connection?
Answer: D. HTTP/HTTPS endpoint with custom URL string
Providing the HTTP or HTTPS endpoint URL is correct because OpenShift AI requires a specific custom endpoint string when connecting to on-premises S3 storage like Ceph. Standard AWS references like ARNs are strictly for cloud resources and will fail locally.
Q10. Which server backend accelerates inference serving payloads for deep learning logic?
Answer: C. Triton Inference Server container runtime engines
The Triton Inference Server is correct because it is the specialized backend integrated with KServe to optimize and accelerate deep learning model payloads. Standard web servers like NGINX cannot process GPU inference workloads or parse machine learning models.
Q11. Which runtime service provides the requisite base architecture supporting Caikit NLP deployments effortlessly on OpenShift AI?
Answer: B. Caikit-TGW Serving Runtime engines
The Caikit-TGIS serving runtime provides the required architecture to deploy Caikit-based NLP models on OpenShift AI. Other options do not natively integrate with the model mesh or OpenShift AI deployment formats.
Q12. What interface URL path provides the deployed OpenShift AI admin management panel UI?
Answer: A. https://rhods-dashboard-redhat-ods-applications.apps.[cluster-domain]
The rhods-dashboard route in the redhat-ods-applications namespace provides the primary OpenShift AI user and administrator interface. The other routes either point to legacy components or do not exist.
Q13. Which control panel within the Red Hat OpenShift AI UI displays enhanced pipeline run chronology and run times?
Answer: C. Pipelines -> Runs panel
The Pipelines Runs tab tracks execution history and run durations for pipelines. Candidates should immediately dismiss unrelated sections like Workbenches or notebook settings when asked about pipeline execution details.
Q14. How do you assign view-only access to a specific component of a Data Science Project for a collaborator?
Answer: B. Edit the Project Access tab in the Dashboard and select Viewer
Editing permissions directly in the dashboard's Project Access tab restricts a collaborator to view-only rights. Cluster-wide roles fail the prompt because they grant broad read access across the entire cluster instead of isolating permissions to a single project component.
Q15. Which system directory stores the accessed storage volume mount points inside an OpenShift AI Red Hat-supplied Jupyter Notebook workbench?
Answer: A. /opt/app-root/src
The /opt/app-root/src directory acts as the default working directory where storage volumes are mounted inside Red Hat Jupyter workbenches. Remembering this specific path is a practical exam cue, as typical Linux workspace or home directories will fail.
Q16. What base container OS image architecture provided by Red Hat is typically used to construct valid custom Jupyter workbenches?
Answer: A. Red Hat Universal Base Image (UBI)
The Red Hat Universal Base Image provides the secure, supported foundation required for building custom OpenShift AI workbenches. Non-Red Hat distributions like Ubuntu or Alpine fail immediately due to lack of enterprise support and compatibility.
Q17. Which serving architecture is optimized for high-scale multi-model density on CPU tiers?
Answer: C. ModelMesh serving architecture stack
ModelMesh is specifically designed for high-scale, multi-model serving optimized for CPU resources. KServe is better suited for single-model deployments or dedicated GPU nodes requiring standalone resources.
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.