Robot Foundation Models Pretraining Scientist

Generalist; Somerville, San Mateo (MA, CA); On-site; USD 120,000 - 180,000; Full time

Job summary

Generalist is seeking a talented individual to help build the core intelligence for robotics through large-scale training of models. You will design pretraining protocols, define model architectures, and develop data strategies across multimodal datasets.

The successful candidate will have experience in training transformer models at scale and a strong foundation in PyTorch. This role promises to push the boundaries of robot intelligence in an innovative team environment.

Qualifications

• Experience with generative models (language, audio, video).
• Comfortable debugging at every layer of the stack.
• Interest in building general-purpose robot intelligence.

Responsibilities

• Design and execute large-scale pretraining runs for models.
• Define model architectures and training curricula.
• Develop scalable data mixtures across petabyte-scale datasets.
• Guide data collection operations and source new datasets.
• Run ablations to understand scaling laws and data quality.
• Collaborate to enhance cluster utilization and reliability.
• Transform raw robotic interaction data into model capabilities.

Skills

Deep experience training large transformer models

Multi-node, multi-GPU distributed training

Scaling laws and optimization dynamics

Strong PyTorch fundamentals

Job description

Generalist is seeking a talented individual to help build the core intelligence for robotics through large-scale training of models. You will design pretraining protocols, define model architectures, and develop data strategies across multimodal datasets.

The successful candidate will have experience in training transformer models at scale and a strong foundation in PyTorch. This role promises to push the boundaries of robot intelligence in an innovative team environment.