Software Engineer, AI Tooling
Mountain View
$160k–$275k a yearHybrid
Job details
- Mountain View
- Hybrid
- $160k–$275k a year (posted by the company)
- Full-time
- AI Tooling
- Posted 1 month ago
Sponsorship history
Past filings by Matx in public government data. Not a promise of sponsorship.
- Filed 1 H-1B LCA in FY2025
Full job description
What MatX is Building
MatX's mission is to make the world’s best AI models run as efficiently as allowed by physics, bringing the world years ahead in AI quality and availability. We build custom silicon for training and running the world's largest AI models, with hardware/software co-design across ISA, RTL, simulator, compiler, and kernels. The AI tooling team sits across that whole stack and owns the agent harnesses and shared workflows that turn general-purpose coding agents into ones maximally useful on each layer.
About the position
Frontier agents are stronger on our Rust and Python than on our Bluespec and SystemVerilog. What context, tools, and checks make an agent trustworthy on a testbench or a timing report, and how do you know? Very few people anywhere have worked through these questions on a chip that has to tape out, with the engineers who sign off on it working next to them. You'd work side-by-side with our hardware engineers, learn how RTL design, design verification (DV), and physical design (PD) work here, codify what's already working, and build what's missing.
What You'll Do Here:
- Close the gap where agents are worst at hardware engineering work. Pair with RTL, DV, and PD engineers while they work, watch where their agents stumble, and identify and ship context, tools, or harness improvements.
- Build tools to address problems that only exist now that agents write code, e.g. a review tool that breaks a large agent-generated diff into reviewable pieces.
- Build the eval loop for our AI tooling by taking bugs out of our repo's history, replaying them against different agent configurations, and using the results to decide what we adopt.
Who You Are:
- Have RTL design, DV, or PD fluency, or will build it fast. You don't need to have taped out a chip, but you'd be working RTL, DV, and PD problems daily and judging to what extent an agent’s answer is accurate.
- Use coding agents heavily and have strong opinions on what you let them do, what you don't, and how you verify their work.
- Built engineering workflows (methodology, flow automation, review tooling, agent harnesses) that other engineers adopted and kept.
- Notice LLM agents asserting something confidently wrong and modify their harness in response.
Bonus Points If You Have:
- Built tooling or automation for chip design flows.
