Senior GPU Infrastructure / AI Install Engineer

About the Role

Lead the hands-on bring-up of an enterprise, on-premises NVIDIA H200 (HGX 8-GPU SXM) environment built on HPE hardware with NVIDIA AI Enterprise. You will take this deployment from physical rack-and-cable installation through firmware validation and cluster initialization to hand off a production-ready environment to the client's AI engineering team.

Responsibilities

• Validate facility readiness (power, cooling/airflow, rack weight, network drops) and execute physical racking and cabling of HPE H200 nodes.
• Flash and configure system firmware, BIOS/iLO, GPU VBIOS, NIC/HCA firmware, IOMMU, large-BAR, and performance profiles.
• Install and optimize the Linux OS, NVIDIA data-center driver, CUDA toolkit, Fabric Manager (NVSwitch), and MOFED GPUDirect/RDMA stack for InfiniBand/RoCE.
• Deploy container runtimes with NVIDIA Container Toolkit, Kubernetes, GPU Operator (and/or Base Command Manager), and configure Run:ai for workload scheduling and fractional-GPU sharing.
• Conduct burn-in testing and cluster validation, including NVLink topology checks, DCGM diagnostics, NCCL multi-GPU/multi-node benchmarks, and thermal/power soak tests.
• Support initial NVIDIA Inference Microservice (NIM) model-serving deployments and deliver a documented, repeatable environment with operational runbooks.

Requirements

• 5+ years of hands-on experience deploying NVIDIA data-center GPUs (Hopper, H100/H200, HGX/SXM architectures).
• Deep technical expertise with NVLink, NVSwitch, Fabric Manager, and high-speed fabrics (InfiniBand or RoCE).
• Advanced Linux system administration skills (Ubuntu/RHEL), driver/kernel troubleshooting, and CUDA stack integration.
• Hands-on containerization and Kubernetes orchestration experience using GPU Operator.
• Proven ability to work independently in regulated enterprise environments with disciplined documentation habits.