Senior Model Engineer - RL & Frontier ML (Palo Alto)

Cursor (Anysphere); Palo Alto (CA); On-site; USD 220,000 - 357,000; Full time

Job summary

SpaceXSI is building Super Intelligence to help humanity understand the universe. We train frontier coding agents and scale RL on real user data to make them increasingly effective.

We're looking for Research Scientists who can drive effective RL or mid-training research in a small-team setting. You'll own ambiguous, hard research problems end to end: forming hypotheses, designing experiments, building the training/eval/data needed to test them, and pushing results into the next model.

Qualifications

• 3+ years training or post-training large neural nets (1B+ params), incl. pretraining, fine-tuning, or RL-based methods (e.g., RLHF, DPO).
• Proficient in production-grade Python, PyTorch, JAX, Rust, or C++

Responsibilities

• Design and run rigorous experiments that turn research ideas into measurable gains in model capability
• Improve data recipes through data quality, deduplication, and mixture experiments
• Define what makes RL data and rewards good, including grader design and preventing reward hacking
• Advance training science, from architecture and scaling experiments to post-training recipes for coding and computer-use agents
• Build evals and targeted tasks that expose model weaknesses and guide where the team invests next
• Partner with engineers to take promising results from small experiments to frontier-scale training runs

Skills

Python

PyTorch

JAX

Rust

C++

Education

MS or PhD in CS/ML/math/physics

Tools

Python

PyTorch

JAX

Rust

C++

Job description

SpaceXSI is building Super Intelligence to help humanity understand the universe. We train frontier coding agents and scale RL on real user data to make them increasingly effective.

We're looking for Research Scientists who can drive effective RL or mid-training research in a small-team setting. You'll own ambiguous, hard research problems end to end: forming hypotheses, designing experiments, building the training/eval/data needed to test them, and pushing results into the next model.