Strategic Finance Manager, Infrastructure and R&DNew$180K–$210K

Fireworks AI
Apply now

The opportunity

This role owns the financial analysis behind our two largest investments: GPU infrastructure and R&D. On the infrastructure side, you'll model how we contract for capacity, how efficiently we use it, and what it costs to serve inference — the drivers that determine gross margin.

What you'll do

  • Own the compute model: Build and maintain the financial model for our GPU fleet — capacity commitments, utilization, cost per GPU-hour — and how all of it flows through to unit economics, gross margin, and cash.

  • Support R&D finance: Serve as a finance partner to Engineering, Product, Security, and IT. Support R&D planning end to end: headcount and hiring plans, tooling and vendor spend, and the operating budget.

  • Support capacity planning: Serve as a source of truth for what capacity we hold, what we've committed to, and what we're actually using; partner with infra on the plan of record.

  • Analyze capacity commitments: Evaluate commitment structures and contract terms with cloud and GPU providers, surface optimization opportunities, and track provider performance and credit refunds.

  • Negotiate and manage vendors: Partner on software, tooling, and infrastructure vendor negotiations and renewals, bringing the analysis that gets us better terms.

  • Get operational: Sit in infra and engineering planning, learn how utilization, scheduling, and roadmaps actually behave, and bring that reality back into the financial model.

What they're looking for

  • –7 years of experience beginning with investment banking, private equity, or: equivalent experience — this foundation is required for the modeling and analytical rigor it builds.
  • Ideally, subsequent time in a corporate or startup finance seat (strategic: finance, FP&A, or corporate development) at a technology company.
  • Expert financial modeling skills: you can build from scratch, stress-test your own assumptions, and translate outputs into clear recommendations.
  • Genuine interest in operations. You want to understand the physical and: technical system underneath the numbers, not just consume its outputs.
  • Comfort with messy and imperfect data.
  • Excellent communication with technical and non-technical audiences alike —: you can take a messy question from a founder or an engineer and return a clear one-page answer.
  • Comfortable being both the architect and the operator, with strong instincts: for building structure where none exists and building processes to scale.
  • High ownership and a bias toward action in an environment where priorities shift and ambiguity is the default.