All Practice Exams

100+ Free EX267 Practice Questions

Pass your Red Hat Certified Specialist in OpenShift AI (EX267) exam on the first try — instant access, no signup required.

✓ No registration✓ No credit card✓ No hidden fees✓ Start practicing immediately
~50-60% Pass Rate
100+ Questions
100% Free
1 / 100
Question 1
Score: 0/0

Which Operator must be installed first to manage Red Hat OpenShift AI components on an OpenShift 4.17 cluster?

A
B
C
D
to track
2026 Statistics

Key Facts: EX267 Exam

210/300

Passing Score (70%)

Red Hat

3 hours

Single Section

Red Hat

$400

Exam Fee (USD)

Red Hat

RHOAI 2.13+

Product Version

Red Hat

$130-200K

MLOps Engineer Salary

Glassdoor 2024

3 years

Cert Valid

Red Hat renewal

EX267 is Red Hat's specialist credential for OpenShift AI (RHOAI). It is a 3-hour, hands-on, performance-based exam with a 210/300 (70%) passing score and a $400 fee. The exam covers the RHOAI Operator, DataScienceCluster and DSCInitialization CRs, workbenches, data science pipelines, distributed training with Ray, and KServe/ModelMesh model serving on OpenShift 4. Credential is valid for 3 years and counts toward Red Hat Certified Architect (RHCA).

Sample EX267 Practice Questions

Try these sample questions to test your EX267 exam readiness. Each question includes a detailed explanation. Start the interactive quiz above for the full 100+ question experience with AI tutoring.

1Which Operator must be installed first to manage Red Hat OpenShift AI components on an OpenShift 4.17 cluster?
A.Red Hat OpenShift AI Operator
B.OpenShift Pipelines Operator
C.OpenShift Service Mesh Operator
D.OpenShift Logging Operator
Explanation: The Red Hat OpenShift AI Operator (rhods-operator package) is installed from OperatorHub and provides the controller that manages the DSCInitialization and DataScienceCluster custom resources. Once subscribed, the operator deploys the RHOAI dashboard and reconciles each component (workbenches, pipelines, KServe, ModelMesh, Ray, CodeFlare).
2Which custom resource initializes cluster-wide settings (such as the applications namespace and monitoring) for Red Hat OpenShift AI?
A.DataScienceCluster
B.DSCInitialization
C.OpenShiftAI
D.RHOAIConfig
Explanation: DSCInitialization (DSCI) is the cluster-scoped custom resource that the RHOAI Operator reconciles first. It creates the redhat-ods-applications namespace, configures monitoring, and sets up Service Mesh prerequisites. DSCI must reach a Ready state before the DataScienceCluster CR is reconciled.
3In a DataScienceCluster CR, what value of managementState causes the operator to install and reconcile a component?
A.Enabled
B.Active
C.Managed
D.Reconciled
Explanation: Each component in spec.components of a DataScienceCluster has a managementState field. Managed tells the operator to install and continuously reconcile that component. Removed deletes it; Unmanaged stops reconciliation but leaves resources in place.
4Which namespace does the RHOAI Operator create by default for its application workloads (dashboard, controllers, etc.)?
A.openshift-ai
B.redhat-ods-applications
C.openshift-operators
D.data-science
Explanation: The RHOAI Operator (and its DSCInitialization CR) creates redhat-ods-applications as the default applications namespace where the dashboard, model controllers, and pipeline operators run. The operator itself runs in redhat-ods-operator.
5Where does an administrator install the Red Hat OpenShift AI Operator from in the OpenShift web console?
A.Operators → OperatorHub
B.Workloads → Deployments
C.Networking → Routes
D.Compute → Machine Sets
Explanation: Operators → OperatorHub is where you browse and install operators packaged via OLM. The Red Hat OpenShift AI Operator card appears there; selecting it creates a Subscription that the Operator Lifecycle Manager reconciles.
6An administrator wants to enable KServe single-model serving but keep ModelMesh disabled. Which two managementState values should they set in the DataScienceCluster?
A.kserve: Enabled, modelmeshserving: Disabled
B.kserve: Managed, modelmeshserving: Removed
C.kserve: Active, modelmeshserving: Inactive
D.kserve: On, modelmeshserving: Off
Explanation: In a DataScienceCluster, components.kserve.managementState: Managed installs and reconciles KServe, while components.modelmeshserving.managementState: Removed uninstalls ModelMesh. Managed and Removed are the two values that produce installed/uninstalled outcomes.
7After installing the RHOAI Operator, a DataScienceCluster CR stays in a Pending state. Which condition is most likely the cause?
A.DSCInitialization has not reached a Ready state
B.The cluster is missing the Logging Operator
C.There are no GPU nodes in the cluster
D.The default storage class is set to NFS
Explanation: The RHOAI Operator reconciles DSCInitialization first; it creates the applications namespace and sets up cluster-wide prerequisites. Until DSCI is Ready, DataScienceCluster cannot proceed. Check the DSCI status and operator pod logs in redhat-ods-operator.
8Which command shows the reconciliation status of a DataScienceCluster named default-dsc?
A.oc get datasciencecluster default-dsc -o yaml
B.oc describe rhoai default-dsc
C.oc get rhoai default-dsc
D.oc status dsc default-dsc
Explanation: oc get datasciencecluster default-dsc -o yaml prints the full CR including spec.components and the status section, which lists per-component conditions like Available, ReconcileComplete, and Phase. The short alias is dsc (oc get dsc).
9Which RHOAI component is responsible for the Kubeflow Pipelines control plane within a data science project?
A.datasciencepipelines
B.workbenches
C.trustyai
D.ray
Explanation: The datasciencepipelines component (managed by RHOAI's pipelines operator) provisions a DataSciencePipelinesApplication (DSPA) per project, which deploys the Kubeflow Pipelines API server, persistence agent, scheduled workflow controller, and MariaDB-backed metadata store.
10Which two operators are typically required as KServe dependencies when KServe is set to Managed in RHOAI Serverless mode?
A.Red Hat OpenShift Serverless and Red Hat OpenShift Service Mesh
B.OpenShift Pipelines and OpenShift GitOps
C.OpenShift Logging and OpenShift Monitoring
D.OpenShift Virtualization and OpenShift Data Foundation
Explanation: KServe Serverless mode (the default for KServe in RHOAI) builds on Knative Serving, which is provided by Red Hat OpenShift Serverless, and on Istio, which is provided by Red Hat OpenShift Service Mesh. Both operators must be installed before KServe can serve InferenceServices in Serverless mode.

About the EX267 Exam

Performance-based certification for data scientists, MLOps engineers, and platform engineers operating Red Hat OpenShift AI. EX267 validates hands-on skills in installing the RHOAI Operator, configuring DataScienceCluster and DSCInitialization custom resources, creating data science projects and workbenches, building Kubeflow data science pipelines with Elyra, running distributed training with Ray and CodeFlare, serving models via KServe single-model and ModelMesh multi-model runtimes, and configuring GPU acceleration on Red Hat OpenShift Container Platform 4.

Assessment

Single 3-hour performance-based hands-on section on a live OpenShift cluster with Red Hat OpenShift AI installed

Time Limit

3 hours

Passing Score

210/300 (70%)

Exam Fee

$400 USD (Red Hat)

EX267 Exam Content Outline

15%

Install and Configure Red Hat OpenShift AI

Install the RHOAI Operator from OperatorHub, configure DSCInitialization (DSCI) and DataScienceCluster (DSC) CRs, manage component states (Managed, Removed, Unmanaged), upgrade RHOAI

12%

Manage Data Science Projects and Permissions

Create data science projects (namespaces with RHOAI labels), configure user and group access, RBAC for project members and admins, project quotas

13%

Workbenches and Notebooks

Launch workbenches via the RHOAI dashboard, Notebook custom resource, prebuilt notebook images (Standard Data Science, PyTorch, TensorFlow, CUDA), custom notebook images via ImageStream, persistent volumes for notebooks

13%

Data Science Pipelines

DataSciencePipelinesApplication (DSPA) CR, Kubeflow Pipelines backend, Elyra visual pipeline editor, runtime configurations, S3-compatible storage for artifacts, scheduled pipeline runs

12%

Distributed Training with Ray and CodeFlare

RayCluster and RayJob CRs, CodeFlare SDK, AppWrapper, KubeRay operator, GPU-accelerated training jobs, Ray dashboard

15%

Model Serving with KServe and ModelMesh

KServe single-model serving (InferenceService CR), ModelMesh multi-model serving, ServingRuntime CR (OpenVINO, Triton, TGIS, vLLM, Caikit), model registry, inference endpoints, REST and gRPC

10%

Accelerator Support and GPU Configuration

NVIDIA GPU Operator, Node Feature Discovery (NFD), AcceleratorProfile CR, taints and tolerations, nodeSelector for GPU nodes, MIG partitioning

5%

Object Storage Integration

S3-compatible storage (OpenShift Data Foundation NooBaa, MinIO), DataConnections, secrets for S3 credentials, model artifacts and pipeline storage

5%

Monitoring and Observability

User Workload Monitoring stack (Prometheus, Grafana), metrics for KServe and ModelMesh, ServiceMonitor, distributed-workloads metrics, model performance dashboards

How to Pass the EX267 Exam

What You Need to Know

  • Passing score: 210/300 (70%)
  • Assessment: Single 3-hour performance-based hands-on section on a live OpenShift cluster with Red Hat OpenShift AI installed
  • Time limit: 3 hours
  • Exam fee: $400 USD

Keys to Passing

  • Complete 500+ practice questions
  • Score 80%+ consistently before scheduling
  • Focus on highest-weighted sections
  • Use our AI tutor for tough concepts

EX267 Study Tips from Top Performers

1Stand up a real RHOAI 2.13+ cluster — Red Hat Developer Sandbox includes RHOAI, or install it on a self-managed OCP 4.17+ cluster
2Memorize the DSCInitialization and DataScienceCluster CRD shapes — components like dashboard, workbenches, datasciencepipelines, kserve, modelmeshserving, ray, and codeflare each have managementState: Managed/Removed/Unmanaged
3Drill InferenceService YAML for KServe single-model serving and InferenceService + ServingRuntime for ModelMesh multi-model serving
4Practice creating Notebook CRs and custom notebook ImageStreams; know the prebuilt images (Standard Data Science, PyTorch, TensorFlow, CUDA, TrustyAI)
5Build at least one Elyra pipeline end-to-end and run it on the Kubeflow Pipelines backend via DataSciencePipelinesApplication
6Run a RayCluster and submit a RayJob using the CodeFlare SDK; understand AppWrapper for batch scheduling
7Configure the NVIDIA GPU Operator and Node Feature Discovery, then apply nodeSelector and tolerations so AI pods land on GPU nodes
8Set up S3-compatible storage (NooBaa from OpenShift Data Foundation or MinIO) and create a DataConnection secret for pipeline artifacts and model files
9Enable User Workload Monitoring and inspect metrics for KServe, ModelMesh, and distributed workloads via Prometheus
10Time yourself with full 3-hour mock labs — verification (does the InferenceService actually return predictions?) matters more than just creating resources

Frequently Asked Questions

What is the EX267 pass rate?

Red Hat does not officially publish pass rates. Industry estimates suggest approximately 50-60% of candidates pass on the first attempt because the exam is hands-on and RHOAI is a young, fast-changing product. The passing score is 210/300 (70%). Most candidates need 100-150 hours of practice on a real OpenShift AI cluster before they reliably hit the threshold.

What RHOAI version does EX267 cover?

EX267 currently aligns to Red Hat OpenShift AI 2.13 or later running on Red Hat OpenShift Container Platform 4.17 or later. Always check the official Red Hat exam page before scheduling — RHOAI ships frequent releases and Red Hat updates the exam objectives accordingly. Practice on a current RHOAI release so the dashboard, CRDs, and operator names match what you see on exam day.

How is EX267 different from EX280?

EX280 is the OpenShift platform administration exam (oc CLI, RBAC, routes, SCCs, operators in general). EX267 assumes you already know that and tests AI-specific workloads on top: installing the RHOAI Operator, managing DataScienceCluster and DSCInitialization CRs, workbenches, data science pipelines, KServe and ModelMesh model serving, and Ray distributed training. EX280 holders still need to study the RHOAI-specific surface area for EX267.

What are the EX267 prerequisites?

Red Hat strongly recommends EX280 (OpenShift Administration) before EX267, plus completion of AI267 (Red Hat OpenShift AI) or equivalent experience. There is no hard prerequisite enforced at registration, but the exam assumes solid OpenShift fluency, container concepts, and practical familiarity with Jupyter notebooks, Kubeflow Pipelines, and Python ML frameworks like PyTorch or TensorFlow.

Does EX267 expire?

Yes — EX267 is valid for 3 years from the date you pass. You can recertify by retaking the current EX267, passing a higher-level Red Hat exam, or earning enough Red Hat credentials to maintain Red Hat Certified Architect (RHCA) status. Red Hat sends renewal notifications before expiration so you can plan ahead.

How long should I study for EX267?

Plan for 100-150 hours of hands-on study over 8-12 weeks if you already hold EX280 and have OpenShift experience. If you are new to OpenShift, double that and pass EX280 first. Build a working RHOAI lab (developer sandbox or self-hosted), drill the RHOAI dashboard, CRDs, KServe InferenceServices, and Ray clusters until commands and YAML are muscle memory.

What jobs can I get with EX267?

EX267 qualifies you for: MLOps Engineer ($130,000-180,000), AI Platform Engineer ($140,000-190,000), Machine Learning Engineer ($130,000-200,000), Data Science Platform Lead ($150,000-200,000), and AI Infrastructure Specialist ($135,000-185,000). Demand is concentrated in regulated industries (banking, telecom, healthcare, government) where RHOAI is the supported AI platform on OpenShift.