HPC Systems Engineer (GPU/CPU Infrastructure & Operations)

About the role

The HPC Infrastructure and Operations team is looking for a motivated, technically capable High-Performance Computing Engineer, suitable for early-career engineers. The role supports the deployment, administration, maintenance and optimisation of large-scale GPU/CPU clusters powering AI, machine learning and scientific research.

Key responsibilities

• Operate, monitor and maintain GPU/CPU clusters, including NVIDIA/AMD GPU servers, CPU nodes and enterprise storage; perform diagnostics, firmware upgrades and component replacement.
• Administer Linux-based HPC clusters, Slurm workload management, distributed storage (Lustre or IBM GPFS) and OS patching and configuration.
• Support AI/ML researchers with job execution issues, GPU memory and multi-GPU communication problems, and workload validation and benchmarking.
• Build automation scripts in Python or Bash, maintain documentation and take part in incident response and root-cause analysis.
• Maintain Apptainer (Singularity) container environments and Kubernetes-based AI/ML platforms, including GPU scheduling and container registries.
• Operate and troubleshoot data center switches (Ethernet and InfiniBand), including leaf-spine fabrics, RoCE/RDMA and network performance monitoring.

Required profile

• Bachelor's degree in a technical field; Master's degree or equivalent practical experience preferred.
• 3-5 years of hands-on Linux server administration, hardware diagnostics or infrastructure operations.
• Ability to investigate problems with logs and monitoring tools, and strong analytical, documentation and communication skills.

Required skills

• Linux, Slurm, Lustre or IBM GPFS, TCP/IP, SSH, DNS.
• Preferred: NVIDIA GPU architecture, CUDA, PyTorch, NCCL, multi-GPU environments.
• Preferred: Kubernetes, NVIDIA Container Toolkit, NVIDIA GPU Operator, Apptainer, container registries, Ansible or Terraform.
• Preferred: InfiniBand, RDMA/RoCE, data center switches (NVIDIA Spectrum, Arista EOS, Cisco NX-OS, Juniper).