SMTS Systems Design Eng.

HPC Systems Engineer

The PERSON:

drivers, troubleshooting methods, and system-level debugging

• KEY RESPONSIBILITIES:

• HPC System Administration & Troubleshooting

  Manage and optimize HPC clusters, ensuring high availability and performance.
• Troubleshoot GPU, CPU, network drivers, firmware, and OS-level issues.
• Debug storage, networking, and job scheduling bottlenecks in Slurm-based environments.

• Kubernetes & Cloud HPC Environments

  Deploy and manage HPC workloads in Kubernetes for AI/ML and parallel computing.
• Optimize OpenStack-based HPC clusters with Ceph, Cinder, and Neutron for cloud scalability.
• Implement containerized HPC workflows using Kubernetes and OpenShift.

• Automation & Infrastructure as Code (IaC)

  Develop Ansible and Terraform scripts for provisioning and managing HPC resources.
• Automate job scheduling, cluster monitoring, and log analysis using Python.
• Optimize CI/CD pipelines for HPC and AI/ML applications.

• Performance Tuning & Benchmarking

  Benchmark and optimize multi-node HPC workloads (MPI, NCCL, ROCm, CUDA).
• Tune OS parameters, networking (InfiniBand, RoCE), and Slurm configurations for peak performance.
• Enhance HPC storage performance (Ceph, Lustre, NFS) and distributed computing efficiency.

• Client Support & Collaboration

  Provide real-time technical support and troubleshooting for HPC users.
• Engage with developers, DevOps, and system administrators to optimize cluster performance.
• Document solutions, best practices, and contribute to internal knowledge bases.

PREFERRED QUALIFICATION'S

• Experience with AMD MI300, MI2X0 GPUs, ROCm, MPI, UCX, or XPMEM.
• Exposure to containerized workloads using Singularity or Docker in HPC.
• Familiarity with OpenStack deployment automation (e.g., TripleO, Kolla, or OpenStack-Ansible).
• Experience in customer-facing technical roles, with a strong ability to troubleshoot live issues.