GPU Infra Engineer in London

GPU Infra Engineer in London

London Full-Time On-site
C

About us

Compute is becoming a financial asset class. Hyperscalers will spend over

$700B

on capex in 2026, most of it on

AI infrastructure

, and almost none of it can be hedged, financed, or traded the way any other industrial input can. We're fixing that. Compute Desk builds the market infrastructure for GPU compute: reference price indices on H100, H200, B200 and B300 rental markets

(Compute Signal),

a platform for physical capacity trading

(Compute Trader)

, and the physical delivery and certification layer that connects it all to regulated derivatives markets

(Compute Clear)

. In September 2026 we

partnered with Nodal Exchange

, a CFTC-regulated exchange, with plans to list cash-settled GPU futures that settle against our indices. We're a small team of people who have built exchanges, trading systems, and large-scale infrastructure. We move fast, argue in good faith, and ship things that haven't existed before. The role

Compute Clear only works if we can verify, benchmark, and certify real GPU clusters at scale. That means someone who has actually stood up this hardware, knows where it breaks, and can turn that knowledge into automated systems rather than tribal knowledge. You'll help own our GPU infrastructure function end to end: the validation frameworks we run against delivery partner clusters, the tooling that automates deployment and testing, and become amongst the 1st hires in the team to run it. What you'll do

Design and run cluster validation and certification: performance benchmarks, interconnect testing (NCCL, InfiniBand/RoCE), thermal and power verification, availability monitoring against SLAs

Build automation for cluster deployment, health checks, and continuous testing so certification scales without headcount scaling with it

Set the technical standards for what "deliverable compute" means: node configs, network topologies, storage, cooling envelopes

Work directly with capacity providers (neoclouds, data centre operators) during onboarding, from site walkthroughs to acceptance testing

Feed what you learn on the ground back into our contract specs, index methodology, and product roadmap

Help hire and support the infrastructure engineering lead as we grow

What we're looking for

You've deployed GPU clusters at both ends of the scale: a handful of nodes and hundreds-to-thousands of GPUs, air cooled and liquid cooled (DLC experience is a real plus)

You've dealt with the messy parts: cabling and topology mistakes, firmware mismatches, failing optics, thermal throttling, the difference between a cluster that passes burn-in and one that survives production

You've automated the boring parts: infrastructure-as-code deployments, monitoring and alerting stacks, automated acceptance and regression testing for hardware

Familiarity with the NVIDIA stack: HGX platforms, NVLink/NVSwitch, CUDA-level debugging when needed, NCCL performance tuning

You communicate clearly with non-infrastructure people: traders, lawyers, capacity providers, regulators

Nice to have

Experience on the operator side: neocloud, hyperscaler, HPC centre, or crypto-mining-to-AI conversions

Exposure to benchmarking frameworks (MLPerf, custom acceptance suites) or SRE practice at scale

Any brush with financial markets or an interest in how physical commodities get standardised and traded

Why is this interesting?

Most infrastructure jobs are about keeping the lights on. This one is about defining what a new asset class physically is. The validation standards you help write become the delivery specs behind exchange-traded contracts. The clusters you help certify become collateral banks will lend against. Very few engineers get to work at the point where hardware meets market structure, and fewer get to do it this early. Send your CV and a short note on a product you've shipped that you're proud of, to The Compute Index is an equal opportunity employer. We welcome applications from people of all backgrounds and are happy to discuss reasonable adjustments at any stage of the process.

C

Contact Details:

Compute Desk Recruitment Team