Principal Research Scientist, Reinforcement Learning

What you'll do

• Design simulation environments, digital twins, reward functions, and verifiers for enterprise tasks.
• Develop RL post-training pipelines and multi-turn, tool-using agents, translating research into production.
• Set technical standards, review architecture and code, mentor colleagues, and contribute to publications.

What they ask for

• 7+ years in ML/AI research or engineering, including 3+ years at senior/staff level.
• 5+ years of hands-on RL and at least one production deployment; 3+ years fine-tuning LLMs with RL post-training.
• Expert implementation of RLHF pipelines, Bradley-Terry reward modeling, DPO, and KTO; working knowledge of modern post-training and rollout-serving libraries.
• MS or PhD in Computer Science, Machine Learning, or a related field (or equivalent); strong Python and software engineering skills.

Nice to have

• Publications at NeurIPS, ICML, ICLR, ACL, COLM, or similar venues, or open-source contributions to post-training or agent frameworks.
• Experience in Offline RL, Model-based RL / World Models, Hierarchical RL, synthetic data, simulation, healthcare, finance, logistics, compliance, or distributed GPU training.

Tools

RLHF, DPO, GRPO, PPO, KTO, TRL, veRL, OpenRLHF

AI focus

Post-train LLMs with reinforcement learning; build tool-using agents, simulation environments, training-data pipelines, and reward systems for enterprise AI.