Platform Engineer

Benchstack Ai; San Francisco (CA); On-site; USD 250,000 - 300,000; Full time

Job summary

Benchstack Ai is seeking an Infrastructure Engineer to own and scale the internal data production platform for frontier AI labs in San Francisco. This role places you at the core of a seed-stage startup delivering high-quality, long-horizon RL environments and post-training data for frontier AI labs.

You will work closely with researchers and data teams to ensure reliability, performance, and deep integration with ongoing experiments, while shaping the platform for rapid growth in a small,

Qualifications

• 3 to 7 years of strong software or infrastructure engineering.
• Experience with distributed systems: databases, storage, networking, messaging or orchestration.
• Has owned and scaled an internal platform end to end.
• Fast code comprehension and JVM-level debugging; async/concurrent I/O a plus.
• Enjoys collaborating with research and data teams.

Responsibilities

• Own the internal data production platform.
• Integrate the platform with research and data teams.
• Ensure reliability and scale to meet surging demand.
• Handle core infrastructure and systems work across the stack.

Skills

Distributed systems

Internal platform

JVM debugging

Research collaboration

Async I/O

Job description

Infrastructure Engineer, Seed AI, Frontier Lab Customers, SF, $250k to $300k

A seed stage AI startup building high-quality, long-horizon RL environments and post training data for frontier AI labs. All five US frontier labs are customers, plus leading international labs, with active engagements across them. The company is already cash flow positive.

What makes them different: most data vendors scale by throwing people at the problem. This team treats post training data as a technical and research problem. The long-term vision is to build a model that produces post training data autonomously, so the company scales with talent, not headcount.

The team

Around 10 people, and very strong. The founders come from AWS, where they worked on distributed databases and network infrastructure. The wider team includes engineers and researchers from Google, DeepMind and NVIDIA, a Stanford physics PhD who works with national labs, and a research lead who has contributed to well-known interpretability tooling. They recently released a research benchmark that drew strong interest from frontier labs.

The role

Inbound demand from frontier labs is outpacing what the team can deliver. You'll own the internal data production platform that powers everything, and make it as scalable, reliable and well integrated with the research team as possible.

What you'll own

• Building and scaling the internal data production platform
• Tight integration between the platform and the research and data team
• Reliability and scale to meet surging demand from frontier labs
• Core infrastructure and systems work across the stack

What they're looking for

• 3 to 7 years of strong software or infrastructure engineering
• Deep distributed systems experience: databases, storage, networking, messaging or orchestration
• Has owned and scaled an internal platform end to end
• Fast code comprehension - JVM-level debugging and async / concurrent I/O nice to have
• Enjoys working closely with research and data teams, not siloed at the app layer

Nice to have

• Experience building a data production or internal developer platform
• Background at a strong infra company, AI lab or fast-growing startupOpen source contributions or serious side projects