NVIDIA Certified Professional - AI Operations

Exam domains

• Installation and Deployment31%

  Provision DGX/HGX clusters with Base Command Manager and Mission Control: bring up head and compute nodes, configure cluster networking (in-band, IPMI, RDMA), deploy Slurm, Kubernetes with the NVIDIA GPU Operator (driver, container toolkit, device plugin, DCGM exporter, MIG manager), Run:AI, and DOCA Services, and apply firmware, driver, and software-image updates across node categories.

• Administration23%

  Run day-2 cluster operations across Slurm partitions, Kubernetes nodes, Run:AI projects/departments, and Base Command Manager: manage users and RBAC, configure MIG geometries via the GPU Operator MIG manager (single vs mixed strategy, all-1g.10gb/all-balanced profiles), enable GPU time-slicing or MPS, and operate the AI data center reference architecture end-to-end.

• Workload Management23%

  Deploy training and inference workloads across Slurm (sbatch, GRES, gres:gpu, srun), Kubernetes (nvidia.com/gpu resource limits, MIG slice scheduling, nodeSelectors), and Run:AI projects with quotas and fair-share scheduling; pull and run containerized stacks from NGC, allocate GPUs across teams, and use job statistics and queue tooling to balance throughput against utilization.

• Troubleshooting and Optimization23%

Sources

Questions are grounded in 50 references from official and authoritative materials.

• GPU Operator Automated Software Deployment on Kubernetes Clusters
• Container Device Interface (CDI) and Node Resource Interface (NRI) Plugin Support
• NVIDIA Fabric Manager — NVSwitch Memory Fabric Configuration
• GPU Time-Slicing for Kubernetes Workload Oversubscription

Similar exams

• NVIDIA Certified Professional - AI InfrastructureNVIDIA
• NVIDIA Certified Associate - AI Infrastructure and OperationsNVIDIA
• NVIDIA Certified Professional - AI NetworkingNVIDIA