RL Research Scientist: Agentic Rewards & Behavior

JDA WMS PTE. LTD.; Singapore; Remote; SGD 90,000 - 150,000; Full time

Job summary

JDA WMS Pte Ltd in Singapore is seeking a specialist in Reinforcement Learning to design reward mechanisms and align short-term agent decisions with long-term outcomes that impact customer satisfaction and retention.

Collaborate with infrastructure and engineering teams to turn research ideas into production-tested agent behaviors, and perform offline evaluation and ablation studies using PyTorch or JAX. Experience with RL benchmarks and solid Python coding are essential.

Qualifications

• Reinforcement Learning Expertise: Strong research experience in RL, applying techniques to real-world problems.
• Programming & ML Skills: Proficiency in Python and modern ML/RL frameworks such as PyTorch, JAX, or similar tools.
• Strong Analytical & Research Skills: Ability to critically evaluate metrics, investigate benchmark results, and conduct rigorous experiments.

Responsibilities

• Reward Design & Long-Term Decision Making: Develop reward mechanisms and connect short-term decisions with delayed outcomes like satisfaction and retention.
• User Modeling & Adaptive Agent Behavior: Design systems that understand changing user preferences and uncertainties, determine responses.
• Policy Optimization & Evaluation: Develop offline evaluation methods, strong baselines, and ablation studies to improve agent performance.
• Research & Engineering Collaboration: Work with infrastructure and engineering teams to translate research into production agent behaviors.

Skills

Reinforcement Learning Expertise

Strong Analytical & Research Skills

Tools

Python

PyTorch

JAX

Job description

JDA WMS Pte Ltd in Singapore is seeking a specialist in Reinforcement Learning to design reward mechanisms and align short-term agent decisions with long-term outcomes that impact customer satisfaction and retention.

Collaborate with infrastructure and engineering teams to turn research ideas into production-tested agent behaviors, and perform offline evaluation and ablation studies using PyTorch or JAX. Experience with RL benchmarks and solid Python coding are essential.