engineering
Posted Apr 30, 2025Product Engineer
at Blacksmith
New York City, United StatesOn-site
Requirements
- Our first product helps companies run GitHub Actions substantially faster and cheaper by owning and operating our own global fleet of bare-metal machines rather than renting generic cloud VMs. - Today, we orchestrate tens of millions of Firecracker VMs each month, running CI for 3,000+ companies and hit ~$10M in ARR in less than 2 years.
- Our GTM is led by Jon Boyer, formerly Head of Sales at Zapier. - We’re now extending the same CI infrastructure into a broader platform: running agent sandboxes at scale and building our own background coding agent on top of it.
- What you could work on - Build and evolve Codesmith, our background coding agent — designing the workflows, UI, and infrastructure that let engineers delegate coding tasks to AI and ship faster - Create tools that help engineers and AI agents debug flaky CI jobs: surfacing patterns in test failures, correlating logs with infrastructure signals, and making root cause analysis fast and intuitive - Work closely with design to ship polished, high-performance product surfaces across our analytics and
- - Can work closely with design to build polished, high-performance UIs — and aren’t afraid to go deep on browser performance, rendering bottlenecks, or complex component architecture.
Experience
- - Have 3+ years of software development experience.
Benefits
- Bonus - You’ve dabbled in your own side hustles or start-up projects. - You’re comfortable creating new processes and iterating on them quickly. Compensation and
- benefits - Medical, Vision, and Dental insurance. - Competitive base + equity. - 401K match. - Unlimited PTO. - Annual offsite. - Early-exercise stock options - 12 weeks fully paid parental leave (US)
Additional details
- About Blacksmith - We started by building infrastructure to run CI workloads really fast.
- We’ve more than tripled revenue since the start of 2026.
- - We operate thousands of bare-metal machines across multiple regions, regularly schedule 100k+ vCPUs concurrently, and run a petabyte-scale Ceph cluster that we manage ourselves.