Research Scientist, Post-Training

About the Role

Join a research team at an applied AI research and training data company, focused on understanding how datasets affect model behavior. You will design controlled post-training experiments and turn their results into evidence that informs data development and model improvement.

What You'll Do

• Run controlled supervised fine-tuning and reinforcement learning experiments to measure how datasets affect model performance.
• Build shared evaluation infrastructure and benchmarks that support reproducible findings across models, capabilities, and domains.
• Help develop public evaluations and new data types that advance model capabilities.
• Publish external-facing research, blog posts, and technical reports.
• Use experimental results to guide improvements to data quality in collaboration with internal teams.
• Work with partner research labs to communicate clear findings about data and model performance.

What We're Looking For

• 0 to 4 years of experience building or delivering work in LLM training, post-training, or model evaluation, including experience at a frontier AI lab.
• Familiarity with LLM training and evaluation methods, including SFT and RL-based post-training.
• Experience independently designing, implementing, and analyzing controlled experiments, and drawing actionable insights from imperfect results.
• Software engineering ability to build or modify experimental infrastructure, such as evaluation harnesses, training pipelines, or benchmarks.
• Ability to work across domains such as finance, software engineering, and policy, with a practical, hands-on approach.
• Research experience involving model evaluation, post-training, or data curation is valuable.

Compensation & Benefits

Base salary

$150,000 to $250,000 annually. Visa sponsorship is available.

Location

On-site in San Francisco, California, United States.