Software Engineer, RL Environments
San Francisco • FullTime
Posted 1mo ago
About the job
SpaceXAI is building Super Intelligence to help humanity understand the universe, from supercomputers to Grok models and their suite of applications. The RL Environments team is seeking a Software Engineer to build systems that transform real-world data into high-quality environments and tasks for reinforcement learning. This role involves creating more realistic training environments and owning the platforms and quality layers for trustworthy data production at scale. You will collaborate closely with ML Platform and research teams to convert various data sources into training-ready environments, making high-quality RL data more accessible for model training efforts.
Responsibilities
- Build platforms for creating complex, realistic, and diverse RL environments at scale, supporting end-to-end tasks across knowledge-work domains.
- Design end-to-end factories to transform massive raw data into useful and realistic environments.
- Define and apply consistent quality standards across vendor, tutor, acquired, and synthetic data.
- Create self-serve APIs and tooling to accelerate task and environment development for internal teams and external contributors.
- Partner across research, platform, and external teams to translate model-capability goals into effective training tasks and environments.
Requirements
- Strong software engineering fundamentals.
- Experience with data platforms, developer tools, distributed systems, or ML infrastructure.
- Ability to translate ambiguous quality standards into concrete, automated checks.
- Comfort building repeatable pipelines from messy, heterogeneous data.
- Care about model behavior and ability to translate capability goals into tasks and experiments.
- Ability to collaborate well across disciplines and own open-ended problems end to end.