Research Scientist, Robotics Pre-Training and Data Quality

Highlights

• Drive data quality and mixture optimization for robotics foundation models.
• Hill-climb model performance via improved training recipes and benchmarks.
• Build tools for rapid iteration and verify data quality via robot training.

Ideal candidate

A robotics researcher passionate about bringing lab breakthroughs to robust real-world systems, energized by data quality, benchmarking, and iterative hill-climbing on foundation models.

Requirements

• PhD in a technical field or equivalent practical experience
• Experience with architectures, data sources, and training and inference for generative multimodal models for robotics
• Experience optimizing models and systems including performance tuning, experimentation and debugging
• Experience with robot policy training using imitation learning and pre-training realities
• Experience with simulators and real-world robotic platforms
• Experience with benchmarking, data labeling strategies and data mixture optimization