HPC Engineer
Core
Validating and tuning large-scale NVIDIA GPU clusters connected by InfiniBand before handover, then owning the HPC layer in production for AI training and HPC workloads.
Role type
Senior IC HPC Engineer (GPU Infrastructure)
Builds
Production-ready GPU clusters, fabric, and storage infrastructure for AI startups and enterprises
Domain
High-Performance Computing (HPC) and AI Infrastructure
Deliverable
production ML models
Required skills
NVIDIA GPU infrastructure operations, Slurm workload management, InfiniBand fabric management, performance benchmarking and tuning, Linux systems administration, infrastructure automation
Preferred skills
HGX H100/H200/B200/GB200 systems, UFM and GPUDirect RDMA, Parallel file systems (WEKA, VAST, Lustre), Kubernetes integration, MPI-based HPC applications
Technologies
NVIDIA CUDA, DCGM, InfiniBand, Slurm, NCCL, HPL, Git, Linux, Kubernetes, OpenStack
Responsibilities
Run burn-in, health checks, and benchmarks (NCCL, HPL) to validate clusters before handover; Operate production HPC clusters including GPU drivers, CUDA stack, and InfiniBand fabrics; Tune clusters for AI training and HPC workloads; Build GPU and fabric monitoring to catch faults early; Automate provisioning and validation with infrastructure-as-code; Lead incident response and root cause analysis for GPU and fabric issues.