Scaling Physical AI through Interactive Learning in Simulated Reality
Scaling Physical AI through Interactive Learning
Hybrid · Toronto
Design, train, and scale action-conditioned generative world models for physical AI and robotics. Develop evaluation metrics for physical plausibility and long-horizon stability while fine-tuning models against physics-grounded reward systems.
Hybrid · Toronto
You will own the end-to-end policy stack for robotics, including imitation pre-training and RL post-training. Additionally, you will manage on-robot deployment, safety protocols, and real-world evaluation to bridge the gap between simulation and hardware.
Hybrid · Toronto
You will build and maintain large-scale multimodal data pipelines for video, lidar, and robot trajectories using distributed computing frameworks. Additionally, you will be responsible for data curation, auto-labeling, and ensuring strict provenance and licensing compliance for all training datasets.
Hybrid · Toronto
The intern will own a scoped project in areas such as ML performance, AI infrastructure, or robotics with a mentor and a clear deliverable. They will work directly within the team's production repositories and clusters to ship a meaningful artifact.
