Research Scientist in RL Training

Job Summary

To advance reinforcement learning techniques for training large language models, the full-time Research Scientist in RL Training will work in a hybrid or remote capacity, focusing on implementing RL methods, designing data pipelines, and prototyping training recipes that enhance Snorkel's data-as-a-service offerings.

Key responsibilities

• Research and implement reinforcement learning techniques and translate them into customer-ready data products
• Design and build data pipelines that generate high-quality training signals for RL workflows
• Prototype and iterate on end-to-end RL training recipes to inform data deliveries

Required qualifications

• Deep expertise in reinforcement learning from human or AI feedback and reward modeling
• Experience training or fine-tuning large language models at scale, preferably 30B+
• Strong proficiency in Python and ML frameworks, especially PyTorch and HuggingFace
• Solid software engineering fundamentals for building research prototypes
• Ph.D. in machine learning, reinforcement learning, or a related field strongly preferred