Software Engineer, Platform Systems

ABOUT THE TEAM The Platform Systems team at OpenAI operates at the intersection of cutting-edge AI and large-scale distributed systems. We build the engineering and research infrastructure required to train OpenAI’s flagship models on some of the world’s largest, custom-built supercomputers. Our team develops core model training software and works deep in the stack - spanning collective communication, compute efficiency, parallelism strategies, fault tolerance, failure detection, and observability. The systems we build are foundational to OpenAI’s research velocity, enabling reliable, efficient training at frontier scale. We collaborate closely with researchers across the organization, continuously incorporating learnings from across OpenAI into the evolution of our training platform. ABOUT THE ROLE As a Software Engineer, Platform Systems, you will design and build distributed systems that provide visibility into large-scale training workloads and help operate them reliably at scale. You’ll work on failure detection, tracing, and observability systems that identify slow or faulty nodes, surface performance bottlenecks, and help engineers understand and optimize massive distributed training jobs. This infrastructure is critical to operating OpenAI’s training stack and is actively evolving to support new use cases and increasingly complex workloads. This role sits at the core of our training infrastructure, blending systems engineering, performance analysis, and large-scale debugging. IN THIS ROLE, YOU WILL - Design and build distributed failure detection, tracing, and profiling systems for large-scale AI training jobs - Develop tooling to identify slow, faulty, or misbehaving nodes and provide actionable visibility into system behavior - Improve observability, reliability, and performance across OpenAI’s training platform - Debug and resolve issues in complex, high-throughput distributed systems - Collaborate with systems, infrastructure, and research teams to evolv

Department: Scaling; Team: Systems