Senior Software Engineer, AI Infrastructure

About the role

While much of the AI industry has moved behind closed APIs, proprietary datasets, and "black box" infrastructure, Ai2 remains a lighthouse for Open Science. Founded by the late Paul Allen, we are a non-profit research institute dedicated to building AI for the common good. We don't have a stock price to defend or a walled garden to protect. Instead, we have a mission: to provide the global research community with the transparent, high-performance foundations they need to achieve humanity-enriching breakthroughs. What Makes Us Different: Radical Transparency: We don't just release model weights; we release the data, the training code, and the infrastructure insights. We believe the "how" is just as important as the "what." Mission over Margin: Our "bottom line" is scientific impact. This gives us the unique freedom to prioritize technical elegance, long-term stability, and open-source contributions over quarterly profit targets. The Best of Both Worlds: We operate at the pace and scale of a world-class tech startup but with the intellectual soul of a research lab. The Beaker Ecosystem: We build and operate systems like Beaker to coordinate the simultaneous training of frontier models (like OLMo) across massive GPU clusters. Our job is to ensure that the next great AI breakthrough isn't stalled by a resource bottleneck or a proprietary gatekeeper. Your Next Challenge: At Ai2, we believe that the most important AI breakthroughs should be transparent and accessible. Your challenge is to build the infrastructure that makes this possible. You will bridge the gap between our researchers and our GPU clusters. You will be a senior technical contributor responsible for ensuring that when a researcher submits a job, the software schedules it intelligently and the hardware executes it flawlessly. This involves: Designing for Scale: Designing and scaling our orchestration layer to ensure that the highest value workloads receive GPU time. Operational Excellence: Moving our HPC operations from manual intervention to high-level automation. Performance Engineering: Working directly with researchers to squeeze every bit of performance out of our GPU-accelerated computing environment.