About this role
The pAGI Infra team at OpenAI builds and operates the core systems that drive reliable, efficient large-scale model training and evaluation. Their domain spans distributed training infrastructure, inference and grading platforms, compute scheduling, and specialized research tooling designed to turn experimental ideas into validated models.
An AI Systems Engineer in this group will scale the infrastructure supporting training and evaluation workflows by owning projects from initial design through deployment and operation. Daily work combines distributed systems engineering and performance optimization while partnering directly with researchers to eliminate bottlenecks, improve resource allocation, and accelerate the path from experiment to production.
Ideal candidates bring strong software engineering fundamentals alongside background in machine learning infrastructure or large-scale distributed systems. The position suits self-motivated builders who enjoy tackling open-ended problems, debugging across system boundaries, and leveraging precise measurements to enhance reliability and throughput.
What you will do
- Design and operate infrastructure for large-scale training and evaluation to boost throughput, reliability, and resource efficiency.
- Develop shared inference and grading platforms equipped with automated capacity management and health monitoring.
- Enhance compute scheduling and resource allocation methods to minimize idle GPU time and speed up workload recovery.
- Investigate performance bottlenecks across training, orchestration, and inference layers.
- Construct self-service tooling, observability dashboards, and automated validation procedures to streamline researcher workflows.
What you need
- Demonstrate strong software engineering fundamentals with practical experience in building or operating large-scale distributed systems.
- Bring relevant background in machine learning infrastructure, GPU performance optimization, inference systems, or infrastructure tooling.
- Exhibit high self-motivation with the ability to take full ownership of open-ended technical problems.
- Possess a strong track record of debugging across complex system boundaries using empirical measurements to guide architectural improvements.
Benefits
- Offers Equity
Skills and tech stack
Questions candidates ask
What does the pAGI Infra team build?
The team builds and operates systems that make large-scale model training and evaluation reliable, efficient, and easy to run. Their work encompasses distributed training infrastructure, inference and grading platforms, compute scheduling, and research tooling.
What will I work on in this role?
You will own projects from identifying bottlenecks and designing solutions through deployment and operation. Tasks include building shared grading services, improving resource allocation across workloads, and bringing new training stacks into production.
Where is the role located?
The position is based in San Francisco.
What is the compensation range?
The salary ranges from $266K to $500K and includes offers of equity.