FluidStack
Production Engineer, Compute
San Francisco, CA (Remote)$175k–$300kfulltimemidAdded 2 weeks ago
About this role
Fluidstack seeks a Production Engineer to build automation and observability systems for managing one of the world's largest GPU compute fleets. You'll own end-to-end fleet health, design repair pipelines, qualify new hardware generations, and scale infrastructure that grows by entire sites every few months.
What you'll do
- Build metrics pipelines and alerting to track GPU fleet health across Kubernetes and bare metal at scale
- Develop automation to manage GPU failures from detection through triage, parts management, and return to service
- Design and expand GPU qualification platform including burn-in testing and performance baselining
- Own Redfish and BMC tooling for firmware-level telemetry and fleet-scale log collection
- Migrate live compute across production sites and bring new sites online sustainably
- Ensure reliability and operational discipline for rapidly expanding GPU fleet infrastructure
What they're looking for
- Hardware troubleshooting and understanding of firmware/silicon-level failure modes
- Kubernetes orchestration and bare metal infrastructure management
- Automation and scripting to eliminate manual operational toil
- Observability and metrics pipeline development
- Redfish/BMC and low-level hardware access protocols
- Rapid learning in unfamiliar technical domains
- Systems thinking at hyperscale
- Incident response and operational discipline
Benefits
- Work on civilization-scale AI infrastructure
- Extreme ownership with full autonomy and end-to-end scope
- High-velocity environment pushing technological frontiers
- Based in San Francisco, CA
Opens the official application on the employer’s site. No login required.
FluidStack
FluidStack builds AI infrastructure at scale, developing data centers and warehouse operations designed to handle gigawatt-capacity compute deployment. The company is hiring for warehouse engineers, data center operations specialists, product engineers, and people leaders to support rapid infrastructure expansion across multiple sites.
- Website
- fluidstack.io
Likely interview questions
- Walk us through how you'd design a metrics pipeline to track GPU fleet health across thousands of devices in production. What would you measure, and how would you surface the signal from the noise?
- Describe your experience with hardware diagnostics or firmware-level tooling. How comfortable are you reasoning about failures at the BMC/Redfish level versus just the application layer?