Research Scientist — Agentic AI & RL Environments

Patronus AI; San Francisco (CA); On-site; USD 175,000 - 300,000; Full time

Benefits offered by this job

Competitive salary and equity packages

15 days of paid vacation per annum

Parental & sick leave

Health, dental, and vision insurance

Job summary

Patronus AI, a frontier lab in San Francisco, seeks researchers to lead foundational studies in training, evaluation, and deployment of agentic AI systems. You will work at the intersection of reinforcement learning, simulations, and scalable oversight to shape how frontier models are developed and stress-tested.

Expect to tackle open-ended questions, design experiments, and publish or open-source results while collaborating with engineering and product teams to translate research into

Qualifications

• MS or PhD in Computer Science, ML, statistics, mathematics, or related field.
• Experience conducting independent research in RL, NLP, alignment, or related areas.
• Ability to take open-ended problems from 0→1 and deliver outcomes.
• Strong experimental design and result interpretation skills.
• Experience writing clean, reproducible Python code with ML frameworks.
• Ability to work quickly and independently with high research quality.

Responsibilities

• Own ambitious research projects end-to-end, from problem identification to experiments and production impact.
• Advance research in agent simulation, RL, and scalable oversight, including cognition and reward design.
• Design state-of-the-art simulation environments for frontier agents and evaluation.
• Train models, run ablations, and analyze distillation, COT reasoning, and hyperparameters.
• Develop methods to understand behavior in long-horizon tasks and improve them.
• Run rigorous experiments and publish or open-source results.
• Build reproducible research systems and partner with engineering and product to deploy.
• Contribute to Patronus AI’s research direction and thought leadership.

Skills

Independent research

Reinforcement learning

Python programming

Experimental design

Cross-functional collaboration

Communication skills

Education

MS or PhD in Computer Science or related quantitative field

Tools

PyTorch

TensorFlow

Job description

Patronus AI, a frontier lab in San Francisco, seeks researchers to lead foundational studies in training, evaluation, and deployment of agentic AI systems. You will work at the intersection of reinforcement learning, simulations, and scalable oversight to shape how frontier models are developed and stress-tested.

Expect to tackle open-ended questions, design experiments, and publish or open-source results while collaborating with engineering and product teams to translate research into