Senior AI Systems Administrator

• A strong track record as a Systems Administrator, HPC Engineer or MLOps Engineer, ideally on contract
• Hands-on GPU server deployment and tuning at multi-rack scale
• Strong InfiniBand/RDMA/RoCE skills and solid network design
• Advanced Linux, virtualisation and infrastructure-as-code skills
• The ability to get productive quickly in a small, collaborative team

Useful extras: Lustre, Ceph, Weka or PEAK:AIO storage; Slurm or LSF; Python scripting.

A not-for-profit AI organisation in central Cambridge needs an experienced contractor to build and run its GPU infrastructure. You’ll provision and look after multi-rack GPU clusters used for training and inference, from the hardware and fabric up to orchestration. Researchers and engineers rely on this platform every day.

What you’ll deliver

• Multi-rack GPU servers deployed, tuned and stable under training and inference workloads
• High-performance networking built and maintained (InfiniBand, RDMA, RoCE), plus the switches, routers and firewalls around it
• The estate managed as code with Git and Terraform/OpenTofu, so it’s repeatable and handed over cleanly
• Linux hosts, hypervisors and VMs administered across Ubuntu and RHEL
• Ongoing tuning with the people who use the systems, based on how they actually run them