Startups · AI

Compute Analyst

Applied Compute · San Francisco · On-site

← All jobs
About Applied Compute

Specific Intelligence for Your Business. Backed by Kleiner Perkins and Lux.

About the role

At Applied Compute, compute is our core input that makes our post-training and agent infrastructure possible. As Compute Analyst, you'll run the analyses behind how we source, price, and use the GPU capacity that powers everything we train, serve, and deploy for our customers. You'll work closely with the finance and research teams to turn a fragmented, fast-moving supply market into clear numbers the company can make big decisions on. Found on 1752vc Careers, the job board for startup and VC roles.

What they're looking for

  • 2–5 years in a highly analytical role, such as investment banking, private equity, consulting, strategic finance or FP&A, or cloud/infrastructure analytics
  • Excellent financial modeling skills and native fluency in unit economics, margin structure, and scenario analysis
  • Strong technical skills: comfortable in SQL and Python, and happy to pull and clean your own data
  • Real curiosity about the AI compute market: providers, accelerators, pricing dynamics, and where things are heading
  • Clear communication, with the ability to explain complex analysis simply to finance, engineering, and leadership
  • High ownership, with a habit of seeing gaps and building the fix before anyone asks
More about this role

At Applied Compute, compute is our core input that makes our post-training and agent infrastructure possible. As Compute Analyst, you'll run the analyses behind how we source, price, and use the GPU capacity that powers everything we train, serve, and deploy for our customers. You'll work closely with the finance and research teams to turn a fragmented, fast-moving supply market into clear numbers the company can make big decisions on. This role suits someone who wants to learn the economics of AI infrastructure from the inside, with real ownership from day one.

Build and own the models behind our compute unit economics, including effective $/GPU-hour, cost per training run, cost per token served, utilization, and goodput

Track the global GPU market: pricing, availability, and terms across hyperscalers, neoclouds, regional operators, and emerging providers, plus outline the tradeoffs between hardware generations

Support live negotiations with quote benchmarking, scenario analysis, side-by-side comparisons of term sheets and contract structures, and deal memos that make the recommendation clear

Work with research and engineering to turn training roadmaps, RL workloads, and...

Read the full posting on Applied Compute's site ↗

GTM and Ops

Build your edge while you search

Free tools for founders and investors, plus VC Unfiltered, our take on startups, venture and the people who build them.