engineering
Posted 2 weeks agoPrincipal Software Engineer, Simulation
at openai
San Francisco, United StatesOn-site
Requirements
- The Simulation team sits at the intersection between the agentic harness that powers OpenAI's products and the research infrastructure where GPT-next is trained, ensuring that our model's training environment is as realistic as possible.
- This role is ideal for a senior backend or infrastructure engineer with strong technical judgment, product sense for highly technical users, and the ability to drive execution across multiple teams.
- IN THIS ROLE, YOU WILL - Design, build, and evolve the integration between the Codex harness that powers OpenAI's products and research training infrastructure used for training GPT-next - Build a platform for our LLMs to train and be evaluated in simulated environments that mimic their deployment setting as closely as possible, on every axis: agentic harness, compute substrate, timing, tools, data sources, humans in the loop, and more - Own major integration surfaces end-to-end, from architecture and API
- experience building and scaling backend or infrastructure systems in fast-moving environments - Bring deep strength in API design, systems design, and engineering fundamentals - Are highly detail-oriented and care deeply about correctness, reliability, and operational quality - Can work directly with demanding technical users while maintaining strong engineering discipline - Have a track record of leading cross-functional technical efforts and creating clarity across organizational boundaries - Bring
- experience with backend platform engineering; Rust
- About OpenAI OpenAI is an AI research and deployment company dedicated to ensuring that general-purpose artificial intelligence
Additional details
- ABOUT THE TEAM OpenAI's research training infrastructure powers how our frontier models are trained and evaluated.
- This team owns the integration layer that connects our production harness capabilities into the training stack.
- The work is highly cross-functional and high leverage: researchers depend on it to run experiments and evaluations reliably as well as to develop the next generation of harness capabilities.
- Failures in this surface can materially affect training velocity and correctness.
- You'll own a critical interface between research and engineering, building the systems, APIs, and operational patterns that let researchers use agentic coding infrastructure safely and effectively in training environments.
- The highest-leverage work is building robust infrastructure that supports and accelerates research without compromising engineering quality.
- experience is a plus LOCATION This role is ideally based in San Francisco due to the close collaboration required with researchers and applied engineering partners.