Applied Scientist: Real-Time Optimization & RL

Amazon; Boston (MA); On-site; USD 143,000 - 193,000; Full time

Benefits offered by this job

Health insurance

401(k) matching

Paid time off

Parental leave

Job summary

Amazon.com Services LLC seeks an Applied Scientist to advance optimization, RL, and sequential decision-making in production environments. You will design and deploy novel decision policies and evaluation methods, collaborating with science and software teams to scale impact across fulfillment operations.

The role emphasizes rigorous evaluation, safe offline training, and translation of research into production planning and control systems, with a strong product focus.

Qualifications

• PhD or equivalent in a quantitative field.
• 2+ years building ML models or algorithms for business applications.
• Experience with optimization methods (LP/nonlinear).
• Proficiency in Python for scripting and ML workflows.
• Experience with end-to-end ownership of major project deliverables.
• Familiarity with ML/RL tooling (PyTorch, RL libraries).
• Experience applying optimization/RL to real-world control problems.

Responsibilities

• Own research and development of optimization and sequential decision-making solutions for real-time MHE control and scheduling.
• Formulate production scheduling problems and design multi-objective functions.
• Build high-fidelity simulation environments for offline training and validation.
• Collaborate across science and engineering teams to integrate policies into production systems.
• Communicate results clearly to technical and business audiences and publish when merited.

Skills

Optimization math

Reinforcement learning

Python

End-to-end ownership

Education

PhD in Operations Research / Statistics / Applied Mathematics / Engineering / CS

Tools

PyTorch

d3rlpy

Ray/RLlib

Gymnasium

Stable-Baselines3

Isaac Gym/Omniverse

Job description

Amazon.com Services LLC seeks an Applied Scientist to advance optimization, RL, and sequential decision-making in production environments. You will design and deploy novel decision policies and evaluation methods, collaborating with science and software teams to scale impact across fulfillment operations.

The role emphasizes rigorous evaluation, safe offline training, and translation of research into production planning and control systems, with a strong product focus.