AI Infrastructure Engineer

• Hands-on experience supporting NVIDIA InfiniBand, preferably the Quantum series, within an HPC or GPU cluster environment. \n

• Experience with high-speed Ethernet switching in a multi-node GPU environment. \n

• Familiarity with NVIDIA HGX architecture and rail-aligned GPU networking topologies. \n

• Experience using NCCL, HPL, or comparable tools to validate GPU cluster health, connectivity, and performance. \n

• Working knowledge of the NVIDIA GPU Operator, GPU driver and firmware lifecycle management, and compatibility planning. \n

• Experience deploying and supporting container-based workloads in an HPC or AI environment. \n

• Experience with at least one workload orchestration platform, such as Slurm, Run, or an equivalent scheduler. \n

• Strong knowledge of large-scale IP networking and datacenter peering/interconnect technologies. \n

• Experience working within structured change-management and production operations processes. \n

• Ability to troubleshoot across physical infrastructure, network fabric, operating systems, containers, orchestration platforms, and application workloads. \n, * Experience administering NVIDIA UFM, NetQ, or equivalent InfiniBand fabric-management tools. \n

• Experience with new GPU cluster bring-up or greenfield AI infrastructure deployments. \n

• Exposure to MLOps pipelines, production AI workloads, and model-serving infrastructure. \n

• Familiarity with DWDM and wave circuit technologies in a datacenter interconnect environment. \n

• Experience supporting enterprise GPU infrastructure in a regulated or highly controlled environment. \n, The strongest candidate will combine deep GPU-cluster networking knowledge with practical platform engineering experience. They should be comfortable moving between InfiniBand troubleshooting, Ethernet and datacenter operations, GPU software lifecycle management, workload scheduling, and production AI enablement.

n \n

• Manage the lifecycle of NVIDIA AI Enterprise and the NVIDIA GPU Operator. \n

• Maintain driver, firmware, CUDA, and platform compatibility across the cluster. \n

• Administer Run, including quotas, project structures, resource allocation, and tenant onboarding. \n

• Configure and support Slurm across dedicated worker-node pools for scheduled batch workloads. \n

• Manage container images, runtime dependencies, and model-serving components across production and pre-production environments. \n

• Partner with the DNA MLOps team to onboard the first production workloads onto the new cluster. \n

• Support engineering and data science teams as adoption of the GPU platform expands. \n

\n

\n