HPC Systems Administrator – High‑Performance Compute Specialist

About the role

Accenture is building a high‑performance compute operations team in London to support cutting‑edge AI infrastructure for sovereign AI projects. The role involves designing, deploying and managing GPU‑based HPC clusters while ensuring secure, reliable service for AI model training.

Key responsibilities

• Design, deploy and manage HPC infrastructures, including GPU clusters and parallel computing environments.
• Support AI model training platforms by maintaining compute resources, optimizing scheduling and ensuring compatibility with AI frameworks.
• Monitor, analyse and fine‑tune performance metrics, addressing bottlenecks and inefficiencies.
• Develop and maintain automation scripts and tools (PowerShell, Python, Bash) for operational tasks, monitoring and reporting.
• Document architecture, configurations, processes and resolutions for compliance and knowledge transfer.
• Participate in root‑cause analysis and post‑incident reviews, implementing preventive measures.

Required profile

• Proven expertise in HPC environments, GPU cluster administration and workload schedulers (SLURM, PBS).
• Experience supporting AI/ML frameworks such as TensorFlow, PyTorch and CUDA.
• Strong understanding of networking, storage and server platforms in both Windows and Linux.
• Advanced analytical, troubleshooting and performance‑tuning abilities.
• Experience with automation, monitoring platforms and scripting languages.
• Excellent communication skills and ability to work across technical and non‑technical teams.
• Familiarity with compliance, data security and relevant certifications (ITIL, NVIDIA DLI, Dell EMC).

Required skills

• NVIDIA and AMD GPU administration
• SLURM and PBS workload schedulers
• TensorFlow, PyTorch, CUDA
• Windows and Linux server environments
• PowerShell, Python, Bash scripting
• ITIL, NVIDIA DLI, Dell EMC certifications

What we offer

• Competitive salary and benefits package
• Opportunities to work on next‑generation AI compute infrastructure
• Travel to client sites across the UK on an ad‑hoc basis
• Security‑cleared environment (BPSS and SC required)