Research Engineer: Open-Ended RL & Data Systems

Proximal; San Francisco, Northern (CA, KY); Hybrid; USD 150,000 - 230,000; Full time

Job summary

Proximal is building research systems to identify limitations of frontier models, create tasks to teach them, and measure improvements with data and evaluations for frontier AI labs.

As a research engineer, you will tackle open-ended problems across RL, agents, and data, sharing work through open-source releases, papers, and talks. You will own problems end-to-end, blending research judgment with strong engineering execution.

Qualifications

• Strong engineering judgment in open-ended research problems.
• Ability to formulate rigorous experiments in ambiguous settings.
• Creative problem solving with strong experimental instincts.
• Self-directed with ownership of problems end-to-end.
• Delivers quality work quickly without sacrificing reliability.

Responsibilities

• Design experiments to identify gaps in frontier models across coding tasks.
• Develop methods to automatically generate harder tasks as capabilities evolve.
• Measure quality of agent-generated code beyond just functional correctness.
• Modify RL training stack to support end-to-end training with subagents.
• Train models to adversarially stress-test data for reward hacking, fairness, and quality failures.

Skills

Engineering judgment

Ambiguity tolerance

Experimental mindset

Self-directed

Execution speed

Job description

Proximal is building research systems to identify limitations of frontier models, create tasks to teach them, and measure improvements with data and evaluations for frontier AI labs.

As a research engineer, you will tackle open-ended problems across RL, agents, and data, sharing work through open-source releases, papers, and talks. You will own problems end-to-end, blending research judgment with strong engineering execution.