Remote Senior HPC Cluster Architect for DL Compute & Storage

NVIDIA Corporation; Warszawa; On-site; PLN 221,250 - 507,000; Full time

Job summary

NVIDIA Corporation in Poland (Warsaw) is seeking a Senior HPC Cluster Administrator to lead design, deployment, and reliability of our large GPU compute clusters across DGX, HGX, and Grace platforms. You will own the full lifecycle, design storage strategies, automate with Ansible and Terraform, manage Slurm scheduling, and monitor systems with Prometheus and Grafana.

Collaboration with ML teams to optimize distributed training is essential.

Qualifications

• BS/MS in CS, EE, CE, or equivalent hands-on experience.
• 5+ years deploying and administering large-scale HPC or ML training clusters.
• Strong Linux systems administration at scale.
• Scripting in Python and/or Bash.
• Experience with Slurm scheduling and accounting.
• Configuration management and IaC tools (Ansible required).
• Experience with containers (Docker, Apptainer/Singularity, Kubernetes).
• High-speed networking (InfiniBand, RoCE, RDMA, EFA).
• Distributed/parallel filesystems and storage architectures.

Responsibilities

• Own the full lifecycle of GPU compute clusters — procurement, provisioning, configuration, monitoring, and deprecation.
• Design and scale storage solutions with capacity and performance roadmaps.
• Lead infrastructure automation using Ansible, Terraform and CI/CD pipelines.
• Manage and optimize Slurm scheduling, fair-share policies, and MIG/GPU partitioning.
• Maintain observability stacks (Prometheus, Grafana, DCGM) and resolve incidents.
• Collaborate with ML engineers to tune cluster configs for large-scale training.
• Evaluate new tech to improve performance and reliability.

Skills

Linux administration

Python

bash scripting

Slurm

Ansible

Terraform

Docker

Kubernetes

InfiniBand

RoCE RDMA EFA

Education

BS/MS in CS/EE/CE

Tools

Ansible

Terraform

Job description

NVIDIA Corporation in Poland (Warsaw) is seeking a Senior HPC Cluster Administrator to lead design, deployment, and reliability of our large GPU compute clusters across DGX, HGX, and Grace platforms. You will own the full lifecycle, design storage strategies, automate with Ansible and Terraform, manage Slurm scheduling, and monitor systems with Prometheus and Grafana.

Collaboration with ML teams to optimize distributed training is essential.