Research Scientist, Post-Training

About The Role

We're looking for a Research Scientist to lead work on post-training data curation for foundation models. You'll design and implement algorithms to generate and improve instruction, preference, and other post-training datasets. You'll also help bridge the gap between pre-training and post-training by exploring how to jointly optimize data across stages.

What You'll Work On

• Post-training data curation: Conduct research on how to algorithmically curate post-training data—e.g., how to generate and refine preference and instruction-following data, how to curate capability- and domain-specific data, and make post-training more effective, controllable, and generalizable.
• Unifying pre-training and post-training data curation: Pursue research on end-to-end data curation: how to curate pre-training data to improve the post-trainability of models and how to jointly optimize pre- and post-training data curation, all in service of maximizing the final performance of post-trained models.
• Transform messy literature into practical improvements: Source, vet, implement, and improve promising ideas from the literature and of your own creation.
• Conduct science driven by real-world needs: Your research will be guided by concrete customer needs and product improvements.

About You

• 3+ years of deep learning research experience
• Experience with post-training large vision, language, and multimodal models
• Post-training algorithm development, data curation, and/or synthetic data methods for: Preference-based tuning (e.g. DPO, RLVR, RRHF), Alternative supervision & self-supervision techniques such as self-training and chain-of-thought distillation, SFT (e.g. instruction tuning and demonstration fine-tuning)
• Post-training tooling development and engineering experience
• Strong understanding of the fundamentals of deep learning
• Sufficient software engineering + deep learning framework (PyTorch or a willingness to learn PyTorch) skills to conduct large-scale research experiments and build production prototypes
• Demonstrated track record of success in deep learning research, whether papers, tools, or other research artifacts

Additional Considerations

• Experience with data management and distributed data processing solutions (e.g. Spark, Snowflake, etc.)
• Experience building + shipping ML products

Compensation

The salary for this position ranges from $180,000 to $300,000. Starting pay is based on job-related skills, experience, qualifications, and interview performance. Our benefits are built to support your well-being and growth:

• 100% covered health benefits (medical, vision, and dental)
• 401(k) plan with a generous 4% company match
• Unlimited PTO policy
• Paid Parental Leave of 12 weeks, plus 6 months of WFH flexibility
• Annual $2,000 wellness stipend
• Annual $1,000 learning and development stipend
• Daily lunches and snacks are provided in our office!
• Relocation assistance for employees moving to the Bay Area.