Founding GPU Infrastructure Lead

TechShack; España; Remote; EUR 85,000 - 120,000; Full time

Job summary

TechShack is hiring for a first-infrastructure role to set technical direction for cluster validation, qualification, and health. You will report to the founders and shape how validation is performed across EU remote work.

We prioritise hands-on leadership with deep expertise in NVIDIA hardware, topology design, and scalable telemetry. You will drive repeatable validation processes and collaborate with a small, funded team across the EU.

Qualifications

• GPU cloud or hyperscaler experience required.
• Experience with large-scale NVIDIA deployments (10,000+ GPUs).
• Strong opinions on topology and failure reasons with supporting data.
• Senior technical voice, no immediate people management.

Responsibilities

• Cluster bring-up and qualification on current-generation NVIDIA hardware across deployments.
• Fleet health and telemetry across diverse hardware.
• Auditing third-party data centre builds to identify issues before deployment.
• Deciding whether a site delivers the specification it sells.
• Developing automated validation processes to make validation repeatable.

Skills

GPU compute experience

NVIDIA hardware

Topology design

Telemetry & fleet health

Cloud infrastructure

Senior technical voice

Tools

NCCL

Telemetry tools

Job description

First infrastructure hire. You set the technical direction for how clusters get validated, qualified and kept healthy, and you report to the founders while you do it.

What you get to work on

• Cluster bring-up and qualification on current-generation NVIDIA hardware, from small deployments to 10,000+ GPU estates
• Fleet health and telemetry across hardware you didn't build, so far more variety than running one estate
• Auditing third-party data centre builds: fabric design, rack layout, power density, failure domains. You find what's wrong before the money moves
• Deciding whether a site actually delivers the specification it's selling
• Building the automation behind the clearing side, so validation is repeatable rather than heroic
• Topology design and on-site assessment as the remit grows

What we need

• GPU cloud, neocloud or AI compute provider background. Hyperscaler-only tends not to translate, as the roles are cut too narrow
• Clusters taken live by you personally, across more than one NVIDIA generation. Racking, burn-in, NCCL runs, RMA chasing
• Experience at both ends of the scale. A 10,000 GPU estate fails differently to a 500 GPU one, and we need someone who has seen both
• Strong opinions on topology and why builds fail, and the evidence to back them
• Senior technical voice without line management. Management available later if you want it

The setup

Greenfield. No inherited estate, no legacy standards to work around, you define how validation gets done. Small, well-funded team, fully remote across the EU. Founding equity, real numbers on the first call. No trading or quant background needed.