AI Researcher - Robotics

We’re seeking a AI Researcher with a strong background in embodied AI, robotics learning, or multi-modal generalization to define what data matters, how to collect it, and how to use it to train models that can generalize across environments and tasks.

You’ll play a pivotal role in:

• Designing the learning objectives and model architectures
• Deciding which modalities and behaviors to capture in the field
• Architecting the data→model→robotics pipeline

• Define data collection strategy to support generalization in robotic models (e.g., visual imitation, action prediction, cross-task transfer)
• Research and implement state-of-the-art models for multi-modal learning (e.g. vision + IMU + gaze)
• Prototype and evaluate model architectures (transformers, diffusion, imitation, contrastive learning, etc.)
• Design pretraining objectives, reward signals, and representation learning tasks

• Deep experience in machine learning, deep learning, and embodied AI (e.g., 2+ years in a top lab or company)
• Strong research track record in one or more of:
• Imitation learning, reinforcement learning, behavioral cloning
• Egocentric perception (Ego4D, EPIC-KITCHENS)
• Robotics policy learning (RT-1, Gato, VIMA, etc.)
• Proficiency in PyTorch, Python, and relevant ML toolkits
• Familiarity with real-world data noise and model robustness

• Publications in ICLR, NeurIPS, CoRL, CVPR, RSS, ICRA
• Experience working with datasets like RoboNet, RT-1, Ego4D, BEHAVIOR, or Habitat
• Exposure to robotic control or sim2real transfer
• Familiarity with gaze estimation, hand-object interaction modeling, or activity segmentation