Jobs / Luxoft

AI Platform Engineer

Luxoft · London, ENG, United Kingdom
London, ENG, United KingdomOnsite
Remuneration
Not specified
Location
London, ENG, United Kingdom
Visa sponsorship
Not specified

Job summary

Engineer and support an enterprise AI platform running Red Hat OpenShift and OpenShift AI on NVIDIA GPU infrastructure, supporting AI/ML workloads on single-node and distributed systems. Manage GPU resources and accelerated networking from bare-metal worker nodes through platform services, using technologies including CUDA, DCGM, SR-IOV, Spectrum-X, RoCE, RDMA and NCCL. Test performance and resilience, troubleshoot infrastructure issues, and collaborate with technology vendors.

Qualifications

  • Experience engineering or supporting Red Hat OpenShift AI and AI/ML workloads.
  • Hands-on experience with NVIDIA GPU infrastructure, GPU Operator, drivers and CUDA.
  • Understanding of GPU scheduling, monitoring and utilisation, including DCGM.
  • Knowledge of NVIDIA Spectrum-X or similar high-performance networking.
  • Experience with the NVIDIA Network Operator, SR-IOV and secondary networks.
  • Understanding of RoCE, RDMA and GPUDirect RDMA for distributed GPU workloads.
  • Experience with NCCL testing and multi-node GPU communication.
  • Understanding of GPUs, NICs or HCAs, PCIe, NUMA, NVLink or NVSwitch, and network fabrics.
  • Experience with enterprise GPU hardware and bare-metal node lifecycle.
  • Experience testing AI platform performance and resilience.
  • Experience working with technology vendors such as Red Hat, NVIDIA and Dell.

Responsibilities

  • Engineer and support Red Hat OpenShift AI and AI/ML workloads on OpenShift.
  • Manage NVIDIA GPU infrastructure, including GPU Operator, drivers, CUDA and GPU resources.
  • Configure and monitor GPU scheduling, allocation and utilisation with NVIDIA DCGM.
  • Deploy and support the NVIDIA Network Operator, SR-IOV and secondary Kubernetes networks.
  • Support high-performance networking using Spectrum-X, RoCE, RDMA and GPUDirect RDMA.
  • Run NCCL tests to validate and troubleshoot multi-node GPU communication.
  • Manage bare-metal GPU worker node lifecycle and hardware, firmware and driver compatibility.
  • Test performance, resilience and failure scenarios for single-node and distributed workloads.
  • Troubleshoot issues across OpenShift, GPUs, networking, storage and physical infrastructure.
  • Collaborate with Red Hat, NVIDIA and Dell to diagnose complex issues.

Skills

KubernetesOpenShiftRHEL

Languages

English: C1 Advanced

Industry

AI/ML

Relocation

No