Fluidstack
Software Engineer, GPU Infrastructure
$208K–$269K
San Francisco, CA +3 · On-site · Full-time · Posted Jul 20, 2026
- Location
- San Francisco, CA
New York, NY
Austin, TX
Seattle, WA
About the job
Build and operate software that monitors, qualifies, repairs, and improves the reliability of Fluidstack’s large-scale GPU compute fleet.
What you'll do
- Build fleet health metrics, alerting, and observability across Kubernetes-orchestrated workloads and bare metal.
- Automate GPU failure detection, triage, parts management, repair, and return to service.
- Develop GPU qualification tools and own Redfish/BMC tooling for telemetry and log collection.
What they ask for
- Experience shipping production automation used by other teams.
- Comfort reasoning about hardware failure modes at firmware and silicon level.
- Ability to operate on call, lead incidents, write postmortems, and address systemic causes.
- Fluency with AI tooling, including LLM APIs, MCP servers, agentic frameworks, and daily use of Claude Code, Cursor, or similar.
Nice to have
- Hardware lifecycle management, RMA automation, BMC/Redfish or IPMI, or GPU qualification/burn-in frameworks.
- Temporal, Cadence, Prometheus, Grafana, Go, or Python.
Pay
$208K – $269K USD total compensation
Tools
Kubernetes, Redfish, BMC, IPMI, Temporal, Cadence, Prometheus, Grafana
AI focus
Use AI coding tools and work with LLM APIs, MCP servers, and agentic frameworks; build infrastructure for AI compute.
Model-written summary · the original posting is authoritative