Compute Lead

Goodfire•San Francisco, CA

7d•$200,000 - $400,000•Onsite

About The Position

We’re hiring a Compute Lead to own one of the most critical resources at the company: our compute for model training, research, and customers. You’ll run everything from how we allocate our compute to how we source, test, and secure capacity. You’ll make the day-to-day calls that determine how far our world-class research and product ambitions can go. We’re looking for someone who can navigate GPU scarcity and partner negotiations with rigor, and moves with exceptional urgency. You’ll need strong judgment to balance competing demands across teams and the conviction to make hard trade-offs rather than split the difference. You might be the right fit if you find the intersection of infrastructure, operations, and high-stakes resource allocation exhilarating.

Requirements

Navigate GPU scarcity and partner negotiations with rigor
Moves with exceptional urgency
Strong judgment to balance competing demands across teams
Conviction to make hard trade-offs rather than split the difference
Find the intersection of infrastructure, operations, and high-stakes resource allocation exhilarating

Responsibilities

Manage Goodfire’s compute allocations across training, research, and customer workloads
Source and secure capacity: gather availability and pricing across providers, move quickly on the best options, and take deals from first call through contract and delivery
Own cluster acceptance testing: define acceptance criteria with our infrastructure team and run burn-in and performance testing before we sign off
Monitor cluster health once live: track node failures and downtime, and drive fixes with providers
Track utilization and quota usage by project, flag idle or underused capacity, and reallocate quickly so GPUs are going to our highest-priority work
Partner closely with research and product teams to translate roadmaps into concrete compute requirements and ensure infrastructure never becomes a bottleneck
Build forecasting models that anticipate compute demand and enable confident long-term planning and budgeting