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: Which graphical interface tool allows teams to assemble pipelines graphically via simple drag-and-drop designs?. 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
- Which graphical interface tool allows teams to assemble pipelines graphically via simple drag-and-drop design…
- Which metric checks whether a model's prediction accuracy drops because the relationship between input featur…
- Which protocol is used by the client application to query a model deployed inside a Triton inference server o…
- Which configuration strategy enables offline RHOAI engines to successfully access model schemas without inter…
- Which open-source toolkit provides the optimization back-end for the OpenVINO Model Server engine?
- Which access mode specification defines persistent volumes requiring simultaneous multiple read and write acc…
Answers and explanations
Tap a question to expand the answer and the exam reasoning. Try to commit to your own pick first.
Q1. Which graphical interface tool allows teams to assemble pipelines graphically via simple drag-and-drop designs?
Answer: A. Elyra Pipeline Editor built into JupyterLab environments
The Elyra Pipeline Editor is a JupyterLab extension that provides a visual drag-and-drop interface for building data pipelines. The OpenShift web console does not offer a visual pipeline builder.
Q2. Which metric checks whether a model's prediction accuracy drops because the relationship between input features and target variables changes?
Answer: B. Concept Drift evaluation metrics
Concept drift metrics detect when the relationship between input features and the target variable changes, degrading prediction accuracy. Covariate drift tracks input distribution changes independently.
Q3. Which protocol is used by the client application to query a model deployed inside a Triton inference server on ModelMesh?
Answer: C. HTTP or gRPC request formats
ModelMesh uses standard HTTP or gRPC protocols to handle inference requests for Triton and other runtime servers. SSH and messaging protocols are not used for runtime inference calls.
Q4. Which configuration strategy enables offline RHOAI engines to successfully access model schemas without internet reachability?
Answer: D. Mirroring source libraries and notebook base images into a secure private registry within the disconnected network
Air-gapped OpenShift AI environments require mirroring all container images and Python packages to a local private registry. Any external tunneling or proxy bypass defeats the disconnected security model.
Q5. Which open-source toolkit provides the optimization back-end for the OpenVINO Model Server engine?
Answer: C. Intel OpenVINO toolkit framework libraries
The OpenVINO Model Server relies on the Intel OpenVINO toolkit as its optimization backend. TensorFlow and PyTorch frameworks do not provide the backend for this serving engine.
Q6. Which access mode specification defines persistent volumes requiring simultaneous multiple read and write access for shared notebook spaces?
Answer: D. ReadWriteMany (RWX) access mode profile
The ReadWriteMany (RWX) access mode allows multiple pods to mount and write to the same persistent volume simultaneously. If a question mentions shared spaces or concurrent access, expect RWX to be the correct choice.
Q7. Which OpenShift mechanism ensures workbenches allocated high hardware resources reside on appropriate specialized nodes?
Answer: A. Applying specific Taints onto specialized hardware nodes matched with Tolerations on target workloads
Applying taints to specialized nodes and adding matching tolerations to target workloads ensures workbenches schedule on the correct hardware. Resource limits alone do not restrict scheduling to specific nodes.
Q8. Which service account is automatically generated to handle pipeline server internal transactions within a Data Science Project?
Answer: A. ds-pipeline-pipeline-server custom internal service accounts
The ds-pipeline-pipeline-server service account is automatically generated to handle internal transactions. Recognizing these specific default service accounts is crucial for troubleshooting RBAC issues during pipeline execution.
Q9. What reconciliation fault causes system component installations to stall permanently at Red Hat OpenShift AI inception stages?
Answer: D. Missing mandatory operator dependencies like Serverless or Service Mesh before setting up managed components
Missing mandatory operator dependencies like Serverless or Service Mesh cause installations to stall permanently. The operator cannot reconcile components without its foundational dependencies, leading to deployment failures during initialization.
Q10. To grant a specific group of users cluster-wide permissions to create custom workbench images, which cluster role is required?
Answer: B. cluster-admin or custom image management cluster roles
Cluster-admin or custom image management cluster roles are required to grant permissions for creating custom workbench images. Standard project roles lack the necessary privileges to push images to the internal registry.
Q11. WHICH PARAMETER ALLOWS THE ADMINISTRATOR TO CUSTOMIZE AVAILABILITY BY SETTING ALLOCATION RULES SO PODS DISPERSE EVENLY ACROSS MULTIPLE NODES?
Answer: B. podTopologySpreadConstraints inside custom manifests / templates
The podTopologySpreadConstraints parameter inside custom manifests allows administrators to set rules so pods disperse evenly across multiple nodes. Standard limit ranges restrict resource sizes but do not control pod distribution across nodes.
Q12. What happens when a data science pipeline step requires a custom Python library that is not present in the runtime container image?
Answer: B. The specific pipeline task step fails with a python ModuleNotFoundError code
If a required dependency is missing, the Python script immediately crashes with a ModuleNotFoundError, causing the pipeline step to fail. OpenShift AI pipelines do not automatically resolve missing packages; you must explicitly define custom dependencies in your pipeline step container image.
Q13. Which object holds the mapping configurations that connect user accounts to specific RHOAI system access permissions?
Answer: A. RoleBinding or ClusterRoleBinding custom resource manifest definitions
RoleBindings and ClusterRoleBindings are standard Kubernetes objects that explicitly map Role definitions to users or groups, granting them specific access permissions. ConfigMaps and Secrets handle configuration data and authentication credentials, but they do not assign authorization rights.
Q14. What metric is evaluated when TrustyAI compares how closely an AI model's output distribution matches a historical baseline distribution?
Answer: B. Concept Drift or Prediction Drift tracking index scores
TrustyAI evaluates prediction and concept drift to compare model output distributions against historical baselines. Focus on drift metrics rather than statistical bias, which evaluates demographic or outcome inequality rather than distribution shifts over time.
Q15. Which configuration reveals an extended hardware definition to data scientists for assignment selections?
Answer: B. AcceleratorProfile resource containing target node tolerations and selectors
The AcceleratorProfile resource exposes specific hardware definitions, like GPUs, to data scientists for workload selection. Notebook configmaps or storage classes do not manage hardware identifiers or integrate seamlessly with the dashboard's resource allocation menu.
Q16. Which custom resource status field indicates that the Red Hat OpenShift AI Operator setup is fully completed and ready?
Answer: B. status.conditions representing Available and Progressing: False
A healthy operator state is shown when status.conditions indicates Available equals True and Progressing equals False. The OpenShift AI operator relies on standard Kubernetes conditions rather than custom phase fields, meaning Pending or InitializedOnly statuses indicate failures.
Q17. What language construct defines conditional control steps inside Data Science Pipelines SDK manifests?
Answer: C. dsl.Condition statement parameters
The Data Science Pipelines SDK uses the dsl.Condition class to define execution branching logic. Python if/else statements evaluate locally during compilation rather than orchestrating runtime conditional graph logic on the cluster.
Q18. Which metric drives scale to zero behaviors within Knative context serving platforms reasoning single-model actions?
Answer: B. Concurrency or request-per-second concurrency levels going down to zero
Knative Serving primarily scales deployments to zero based on actual incoming request concurrency. Memory limits or node compute utilization rules dictate general pod eviction and scheduling, not the zero-scaling Serving actions.
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.