Research Scientist, Agentic RL & Scalable AI

Goaly; Menlo Park, Northern (CA, KY); Hybrid; USD 150,000 - 210,000; Full time

Benefits offered by this job

Meals and office benefits

Job summary

Goaly is seeking researchers at the intersection of agentic reinforcement learning and production systems. You will identify high-leverage questions, design decisive experiments, and translate results into model improvements and reusable systems.

This role suits candidates completing a PhD or with equivalent original research; you will work closely with Post-Training, RL Systems, Training, Backend & Product engineers to scale ideas within production constraints.

Qualifications

• PhD or equivalent record of original, rigorous research.
• Strong research record in ML/RL/LLMs, agents, or ML systems.
• Excellent Python skills with a modern DL framework (PyTorch/JAX).
• Experimental rigor: define falsifiable questions, measure, control confounders, interpret results.

Responsibilities

• Formulate high-leverage research questions about agentic capability, RL, reward design, and scaling.
• Design experiments, ablations, controls, and evaluations that separate true improvement from noise.
• Implement methods in modern DL frameworks and integrate with training/evaluation systems.
• Build or improve datasets, environments, verifiers, and evaluations for coding, tool use, reasoning, and interaction.
• Analyze trajectories and model behavior; develop failure taxonomies and testable hypotheses.
• Collaborate with Post-Training, RL Systems, Training, and Backend & Product engineers to scale ideas.
• Communicate findings in internal docs and reviews; contribute to papers and open-source releases.
• Help shape the research roadmap by identifying reusable evaluation assets and critical uncertainties.

Skills

PhD in CS/ML/Math

Research track record

Python

Experimental rigor

Engineering ability

Communication

Education

PhD in Computer Science, ML, statistics, mathematics

Tools

PyTorch

JAX

Job description

Goaly is seeking researchers at the intersection of agentic reinforcement learning and production systems. You will identify high-leverage questions, design decisive experiments, and translate results into model improvements and reusable systems.

This role suits candidates completing a PhD or with equivalent original research; you will work closely with Post-Training, RL Systems, Training, Backend & Product engineers to scale ideas within production constraints.