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: Which element defines secured network encryption enforcement along mesh networks managed within single-model serving sys. 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 element defines secured network encryption enforcement along mesh networks managed within single-model…
- What type of volume allocation reduces storage provisioning strategically for experimental scratch workspaces?
- Which component parses incoming inference payload data to check for anomalous request distributions over time?
- Which parameter instructs pipeline artifact storage managers to target encrypted repositories for log storage?
- Which storage engine is strongly recommended for disconnected platforms to act as an internal registry for im…
- Which parameter in a serving runtime configuration file indicates whether the runtime supports executing on G…
Answers and explanations
Tap a question to expand the answer and the exam reasoning. Try to commit to your own pick first.
Q1. Which element defines secured network encryption enforcement along mesh networks managed within single-model serving systems?
Answer: A. PeerAuthentication custom resource configurations requiring STRICT mTLS settings across namespaces
Setting PeerAuthentication to STRICT mTLS enforces mesh network encryption for single-model serving. NetworkPolicies restrict traffic flow but do not handle certificate-based encryption or mutual authentication between pods.
Q2. What type of volume allocation reduces storage provisioning strategically for experimental scratch workspaces?
Answer: C. Ephemeral Storage or emptyDir
Ephemeral storage, such as an emptyDir, is tied to the pod's lifecycle and does not require persistent provisioning, reducing overhead for temporary scratch data. Avoid persistent volumes like S3 or block storage unless the data needs to survive a workspace restart.
Q3. Which component parses incoming inference payload data to check for anomalous request distributions over time?
Answer: D. TrustyAI payload processing modules and analytical services engines
TrustyAI analyses incoming payloads to detect bias and monitor anomalous distributions over time. The Service Mesh handles routing, while KServe manages the actual model deployment, making them distractors.
Q4. Which parameter instructs pipeline artifact storage managers to target encrypted repositories for log storage?
Answer: D. S3 bucket credential secrets passed through pipeline server initialization custom resource specs
Passing S3 bucket credentials via the pipeline server's custom resource configures secure artifact and log storage. While Git layout manifests organize files, they do not handle the actual authentication required for encrypted object repositories.
Q5. Which storage engine is strongly recommended for disconnected platforms to act as an internal registry for images?
Answer: B. Red Hat Quay Registry
Red Hat Quay acts as a highly available internal OCI registry for air-gapped OpenShift environments. While other registries function similarly, Quay is the Red Hat enterprise standard for disconnected deployments.
Q6. Which parameter in a serving runtime configuration file indicates whether the runtime supports executing on GPU hardware?
Answer: C. The resources allocations specifications and environment configurations tags (e.g., ACCELERATORS)
The resource allocations and environment configurations, such as accelerators, indicate GPU support in a serving runtime. For the exam, look for the resource requirements and annotations that explicitly request GPU hardware in the serving runtime YAML.
Q7. Which command outputs the real-time event logs of an OpenShift cluster to diagnose notebook pod scheduling issues?
Answer: C. oc get events -n –watch
The oc get events command with the watch flag outputs real-time cluster events to diagnose pod scheduling issues. Checking operator logs helps troubleshoot the operator itself but misses namespace-level notebook pod failures.
Q8. Which user action can cause an automated culling process to shut down an active workbench pod?
Answer: A. Zero interactive active keyboard code cells processing actions within specified timeout limits
The culling process shuts down workbench pods when there are zero interactive keyboard actions in cells within a specified timeout limit. Closing the browser or running background scripts does not always trigger the idle culling mechanism.
Q9. Which open-source project acts as the model configuration and runtime wrapper inside the KServe architecture?
Answer: A. KFServing / KServe Core
KServe Core acts as the standard wrapper for model configuration and runtime within the KServe architecture. Expect KServe questions whenever single-model deployment or serverless inference comes up on the exam.
Q10. Which file system permission structure is applied to mounted persistent volumes inside OpenShift AI workbenches?
Answer: C. Configured to match the arbitrary non-root user ID assigned by the cluster execution project
Persistent volumes in OpenShift AI are configured to match the arbitrary non-root user ID assigned to the namespace. This ensures the notebook process can read and write files without encountering permission denied errors.
Q11. Which component of OpenShift AI acts as an intermediary dashboard to launch individual Jupyter hubs and track settings?
Answer: B. odh-dashboard pod application
The odh-dashboard acts as the intermediary web application for launching workbenches and managing settings. On the exam, expect to verify this specific pod when troubleshooting central dashboard access or feature availability.
Q12. Which field defines the target storage artifact path location within pipeline metadata definition manifests?
Answer: C. spec.pipelineSpec.tasks.artifacts.storageUri property configurations
Artifact storage paths are defined in the storageUri property of the pipeline specification tasks. ConfigMaps and identity tokens handle global settings or authentication, not specific artifact routing for pipeline tasks.
Q13. Which component handles the translation of high-level prompt templates into system instructions inside an LLM orchestration app?
Answer: A. LangChain, LlamaIndex, or application-level orchestration libraries
Orchestration libraries like LangChain map prompt templates into system instructions. KServe merely serves the model, while Elyra handles pipeline scheduling, so neither performs prompt translation within an LLM application.
Q14. Which object record contains encrypted base64 strings for bucket credentials that define data connections?
Answer: C. Kubernetes Secret with specific annotations
Data connections store endpoint credentials in Kubernetes Secrets annotated specifically for OpenShift AI. While a DataConnection custom resource manages the connection, the actual credentials are stored securely in the Secret.
Q15. Which service discovers hardware types and accelerators natively across hardware rigs without manual label inputs?
Answer: A. Node Feature Discovery Operator
The Node Feature Discovery Operator automatically detects hardware features like GPUs and labels nodes accordingly. While the GPU Operator uses these labels, it does not perform the initial hardware discovery across the cluster.
Q16. What kind of service detects abnormal bias deviation within input data for live served machine learning models?
Answer: A. TrustyAI Service component
The TrustyAI Service is designed to evaluate bias and fairness metrics for live machine learning models. Prometheus only handles numerical system metrics, while KServe strictly manages deployment routing.
Q17. Which custom metric type allows TrustyAI to measure fairness based on user-defined calculations rather than standard algorithms?
Answer: B. Custom Bias Evaluators or user-defined metric calculation configurations
Custom Bias Evaluators allow TrustyAI to measure fairness based on user-defined calculations rather than relying solely on standard algorithms. Data drift tracks input changes, not specific custom fairness metrics.
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.