Research Lead, Tinker, Fine-tuning Science
Core
Lead the Fine-tuning Science team to set the research agenda for frontier customization techniques and the Tinker post-training engine, ensuring findings ship into production.
Role type
Senior IC research lead (fine-tuning science)
Builds
Tinker, the leading post-training engine for customizing frontier models
Domain
AI / Machine Learning / Large Language Models
Deliverable
production ML models
Required skills
Research leadership, Python, deep learning frameworks (PyTorch, TensorFlow, JAX), distributed training debugging, communication of complex technical concepts
Preferred skills
Probability and statistics, RLHF/RLAIF/preference modeling, human data collection campaign management, alignment research, RL training stability techniques
Technologies
PyTorch, TensorFlow, JAX
Responsibilities
Set the research agenda for Tinker quality, efficiency, and reliability; hire, mentor, and develop researchers; stay hands-on with LoRA and parameter-efficient fine-tuning; improve stability of large-scale fine-tuning and RL runs; represent work externally via papers and blog posts
Seniority
Senior, hands-on IC with team leadership
