Multimodal Learning Researcher (Junior)

We’re looking for a multimodal learning researcher to help shape and build foundation models for physical interaction. This role spans representation learning, multimodal sensor integration, and model evaluation, requiring someone who can move between designing learning methods, training models on real-world data, and measuring how well they generalize across tasks and hardware.

What you’ll do

• Build foundation models for physical interaction, trained on large-scale multimodal sensor data.
• Develop self-supervised representations across tasks and hardware.
• Integrate new sensing modalities with vision and proprioception for robot learning.
• Design evaluations that measure real-world impact.

What we’re looking for

• Master’s or PhD in ML, CS, Robotics or a related field, or equivalent research experience, with a focus on self-supervised, representation or multimodal learning.
• Experience training deep learning models end to end, from data pipelines through evaluation, with strong PyTorch skills.
• Comfortable with messy, real-world sensor data.

Nice to have: experience with time-series or physical sensor data; a robotics background; world models; publications at top ML or robotics venues.