The mission
You will work every day on the racks that power one of France's largest AI clusters.
Mistral's compute footprint is our single largest investment and the foundation of our AI research and engineering. We're hiring Data Center Hardware Engineers (L1-L3) to keep our GPU/CPU clusters healthy, safe and growing — hands-on, in the data center, at the intersection of high-performance computing and physical infrastructure.
You'll join a small, senior team working at a scale few engineers get to experience: GPU-dense racks (H100/H200-class and next-generation platforms), high-speed interconnects, liquid-cooled systems — with new sites under development in France and beyond.
Hardware here is not someone else's problem. As part of a small on-site team, you carry responsibility for it end to end.
Location
On-site at one of our data centers in the Paris region
What you'll do
Diagnose & repair server and cluster hardware: CPUs, memory, drives, NICs, GPUs, PSUs, and interconnect problems (switches, cables, transceivers; Ethernet/InfiniBand).
Intervene safely — power-off/lockout, ESD discipline, pre/post-work checklists, tidy and safe work areas.
Capture everything — photos, serials, test results; open, update and close tickets with notes the next engineer can act on.
Manage parts — receive, label and track inventory; run RMAs and coordinate with vendors.
Help us get better — feed recurring anomalies into monitoring, alerts and SOPs; contribute small scripts (Python/Bash) if you have those skills.
Be a good teammate — clear status updates, honest handovers, and respect for the people depending on the cluster.
About you
Hands-on comfort in data centers or server hardware — you can (or will quickly learn to) work cleanly in racks: rails, cabling, labeling, mounting.
Disciplined and meticulous: checklists aren't bureaucracy to you, they're craftsmanship.
Practical electrical basics: power-off procedures, PPE, short-circuit risk awareness.
Basic Linux (boot checks, logs) — scripting is a plus, not a requirement.
Physically able to work in racks (lift, mount, maneuver) in a data center environment.
Comfortable in a 24/7 operation, including a rotating on-call with emergency response expectations.
Driver's licence (B) for travel between our sites.
Even if you don't match every criterion
Many of our best hardware engineers came from research labs, telecoms, IT services or the military. If you are rigorous, curious and genuinely motivated by hardware at scale, we would be glad to hear from you — we assess demonstrated scope rather than keyword matching.
Nice to have
HPC/AI/cloud production experience at scale; large-fleet server install and maintenance.
Networking fundamentals (Ethernet/InfiniBand) beyond basics.
Vendor/RMA tooling, ticketing (Jira/ServiceNow), inventory systems.
Python/Bash automation for checks, capture, inventory or reporting.