Australian PR or citizenship required
GPU Infrastructure Engineer | Sydney and Melbourne | Permanent
If you've spent years keeping Linux running hard in production data centres, and you know GPUs inside out, you're exactly who we're looking for. This is your chance to run the backbone of an Australian AI cloud platform where every decision you make has real impact.
About our client
Our client builds cloud infrastructure for AI workloads right here in Australia. They operate high-density data centres delivering GPU compute power to organisations that need serious computational muscle. It's a business that's growing fast, where the engineering matters and the work you do is visible and valued every single day.
The role
You'll own GPU server infrastructure from bare metal to live workloads. Your days will span provisioning new hardware, managing fleet health across hundreds of systems, and tuning CUDA and NCCL stacks to squeeze maximum performance. You'll build and operate GPU-aware Kubernetes clusters, write automation that makes the platform scale elegantly, and troubleshoot the trickiest problems that span hardware, networks, the operating system and applications. You'll work across hardware lifecycle management, out-of-band remote systems, observability tooling and customer support. Most importantly, you'll collaborate closely with the systems, network and product teams to shape how the platform evolves.
What's on offer
Hybrid flexibility across Sydney and Melbourne CBD locations, genuine scope to grow your career, and a team that values deep technical expertise and real ownership of the platform.
What you'll need to succeed
You'll bring eight or more years running Linux systems at scale in production data centres, with hands-on experience operating NVIDIA GPUs and tuning CUDA workloads. Strong Kubernetes knowledge is essential, as is solid grounding in bare-metal provisioning, hardware management, and high-performance networking like InfiniBand and RDMA. You're comfortable writing Python for automation and infrastructure-as-code, and you can troubleshoot across every layer when things get complex. Experience with multi-node GPU systems and Base Command Manager is a plus. You must be an Australian citizen (security clearance will be required). If this describes you, we'd love to hear from you.