RL Data Software Engineer — Shape Better Models

Speedrun Talent Network; San Francisco (CA); Hybrid; USD 150,000 - 190,000; Full time

Job summary

SpaceXAI in SF seeks a Software Engineer, RL Data to shape training data for Grok models. You will determine what problems agents practice, what worlds they train in, and how they are judged, turning weaknesses into measurable improvements.

You will build systems to test hypotheses, run ablations, and read rollouts to find root causes behind model behavior. Realistic simulations and production-like codebases power this work.

Qualifications

• Strong software engineering fundamentals and production-grade coding experience.
• Experience building data pipelines and testing infrastructures.
• Ability to design and evaluate data for ML models.
• Comfort working with complex backend systems and simulations.

Responsibilities

• Turn capability gaps into verifiable tasks within real codebases.
• Create environments that simulate production apps and data flows.
• Design graders to distinguish good solutions from shortcuts.
• Build automated checks to catch problematic data or behavior.
• Scale task generation from thousands to millions with reliable filters.

Skills

Python

Data pipelines

Test infrastructure

Sandbox environments

Backend services

Job description

SpaceXAI in SF seeks a Software Engineer, RL Data to shape training data for Grok models. You will determine what problems agents practice, what worlds they train in, and how they are judged, turning weaknesses into measurable improvements.

You will build systems to test hypotheses, run ablations, and read rollouts to find root causes behind model behavior. Realistic simulations and production-like codebases power this work.