Principal GenAI RL Gym Scientist

Amazon Science; Sunnyvale (CA); On-site; USD 229,000 - 309,000; Full time

Benefits offered by this job

Health insurance

401(k) matching

Paid time off

Parental leave

RSUs

Job summary

Amazon Science is seeking a Principal Applied Scientist to lead the development of training and evaluation assets for frontier GenAI with LLMs. You will set the technical direction for RL gyms and tooling used by model teams across Amazon.

You will design experiments, build scalable tooling, and mentor junior scientists while collaborating with cross‑functional groups to drive measurable improvements in real-world AI systems.

Qualifications

• PhD in CS/EE/Math or related field; or Master’s with significant applied research.
• Strong hands-on experience building ML models and systems with deep learning.
• Experience with reinforcement learning and RL environments or tool-use systems.
• Experience setting technical direction and leading teams of scientists/engineers.

Responsibilities

• Set technical direction across data collection, sampling, and evaluation for RL gyms.
• Design and lead experiments that yield reliable signals for model improvement.
• Build gym-building tooling and automated quality control at scale.
• Advance the science of gym evaluation and validation for model training.
• Hands-on leadership: contribute high-quality assets and mentor junior staff.
• Collaborate with domain experts and cross-functional teams on high-value tasks.
• Lead scientific reviews and help hire and develop other scientists.

Skills

Java/C++/Python

Deep learning

Reinforcement learning

Technical leadership

Mentorship

Education

PhD in CS/EE/Math or related, or Master’s with applied research

Job description

Amazon Science is seeking a Principal Applied Scientist to lead the development of training and evaluation assets for frontier GenAI with LLMs. You will set the technical direction for RL gyms and tooling used by model teams across Amazon.

You will design experiments, build scalable tooling, and mentor junior scientists while collaborating with cross‑functional groups to drive measurable improvements in real-world AI systems.