Applied Scientist - RL Environments for LLMs

Amazon Inc.; Sunnyvale (CA); On-site; USD 172,000 - 222,000; Full time

Benefits offered by this job

Health insurance

401(k) matching

Paid time off

Parental leave

Job summary

Amazon is seeking an Applied Scientist with a strong deep learning background to help build training environments that teach large language models to reason, use tools, and complete complex, multi-step tasks.

You will design, build, and evaluate reinforcement learning environments (RL gyms) for LLM training, including defining tasks and reward signals, and scaling infrastructure. You will work with model training teams to maximize value and verify environment quality.

Qualifications

• 3+ years of building models for business application experience.
• PhD, or Master's degree and 4+ years of CS, CE, ML or related field experience.
• Experience programming in Java, C++, Python or related language.
• Experience in algorithms and data structures, parsing, numerical optimization, data mining, parallel and distributed computing, high-performance computing.

Responsibilities

• Design, build, and evaluate RL environments for LLM training with realistic reward signals and scalable infrastructure.
• Work with model training teams to identify where environments add value and iterate based on model learning.
• Measure transfer of environments to model capability and verify environment quality and correctness.

Skills

Deep learning

Reinforcement learning

Programming in Java, C++, Python

Model development for business apps

Education

PhD, or MS + 4+ years CS/ML/CE

Tools

Unix/Linux

Job description

Amazon is seeking an Applied Scientist with a strong deep learning background to help build training environments that teach large language models to reason, use tools, and complete complex, multi-step tasks.

You will design, build, and evaluate reinforcement learning environments (RL gyms) for LLM training, including defining tasks and reward signals, and scaling infrastructure. You will work with model training teams to maximize value and verify environment quality.