Research Scientist: Pretraining

Generalist • Somerville, MA

5d • $240,000 - $350,000

About The Position

Generalist is building general intelligence for the physical world through embodied foundation models, initially focused on dexterity. The company advances data, models, and hardware so robots can intelligently interact with the physical world.

As a Research Scientist on Pretraining, you will build the base intelligence layer for robotics by training large-scale robot foundation models on massive multimodal datasets spanning video, proprioception, action traces, language, and more. You will design and run core training efforts to develop general capabilities across embodiments, tasks, and environments. The role is suited to researchers with experience pretraining large-scale machine learning systems or generative foundation models, including multimodal, language, audio, or video models.

Requirements

• Experience with large-scale ML systems at senior/staff (L6+) level.
• Deep experience training large transformer or diffusion models at scale for generative models, such as language, audio, or video models.
• Experience leading or significantly contributing to multi-node, multi-GPU distributed training efforts.
• Experience with scaling laws, optimization dynamics, and large-model failure modes.
• Strong PyTorch fundamentals and comfort debugging at every layer of the stack.
• Care for empirical rigor and rapid iteration.

Responsibilities

• Design and execute large-scale pretraining runs for robot foundation models using transformer- and diffusion-based architectures.
• Define model architectures, objectives, and training curricula across multimodal robotic data, including vision, action, state, and language.
• Develop scalable data mixtures and sampling strategies across petabyte-scale datasets.
• Guide data collection operations toward new directions and source new datasets.
• Run ablations to understand scaling laws, data quality effects, and architecture tradeoffs.
• Collaborate closely with ML Infrastructure and Systems teams to improve cluster utilization, throughput, and reliability.
• Turn raw robotic interaction data into generalizable model capabilities.

What This Job Offers

• Job Type: Full-time
• Career Level: Senior
• Education Level: No Education Listed