ABOUT THE ROLE 

We're looking for a Research Scientist to work on reinforcement learning for training and aligning large language models. This is a foundational research role focused on one of the most consequential open data problems in AI: how to generate the data, reward signals, and training procedures that steer LLM behavior in reliable and generalizable directions — and a core capability that directly differentiates Snorkel's data-as-a-service offering. 

You'll work closely with Snorkel's research, engineering, and delivery teams to advance our RL data capabilities — translating research ideas into the preference datasets, reward models, and RL-ready corpora we produce for frontier AI labs, and contributing to a research agenda that is central to Snorkel's long-term differentiation as a provider of bespoke training data. 

MAIN RESPONSIBILITIES 

• Research and implement reinforcement learning techniques — including GRPO, RLHF, RLAIF, DPO, and reward modeling — and translate them into data products (preference datasets, reward signals, verifiable rewards) that customers can use to train and fine-tune large language models. 
• Design and build data pipelines that generate high-quality training signal for RL workflows, including AI-assisted data annotation and curation data pipelines to improve model generalization to unseen benchmarks . 
• Prototype and iterate on end-to-end RL training recipes that inform what data Snorkel ships as part of its data-as-a-service deliveries. 
• Work closely with research scientists, ML engineers, and delivery teams to translate RL research into customer-ready data products.
• Stay current with the latest developments in large-scale muli-node LLM training, alignment research, and scalable RL methods (on complex environments such as Terminal-Bench), bringing relevant advances into Snorkel's data-as-a-service approach.
• Contribute to Snorkel's research publications and internal knowledge base in RL and model training.

PREFERRED QUALIFICATIONS 

• Deep expertise in reinforcement learning from human or AI feedback, reward modeling and credit attribution ideally with a clear perspective on what data makes these techniques work. 
• Experience training or fine-tuning 30B+ large language models at scale, including familiarity with distributed training infrastructure. 
• Strong proficiency in Python and ML frameworks, especially PyTorch and HuggingFace and hands-on experience with RL frameworks such as Verl and SkyRL. 
• Solid software engineering fundamentals — you can build research prototypes that others can run, extend, and integrate into data production workflows. 
• Familiarity with ML infrastructure and cloud platforms and tools (AWS, GCP, Kubernetes, Slurm, etc.); experience with large-scale RL training pipelines a strong plus. 
• Comfort operating in a high-iteration environment with open-ended research questions and shifting, customer-driven technical constraints. 
• Ph.D. in machine learning, reinforcement learning, or a related field strongly preferred; exceptional industry experience considered.