Conduct research on world modeling and 3D geometry, developing self-supervised learning methods, video-based world-dynamics architectures, scalable data pipelines, evaluation frameworks, and model-based planning algorithms. The role requires independent experimentation and analysis in machine learning, large-scale training, and GPU or TPU environments, with preferred expertise in 3D/4D reconstruction, SLAM, SfM, depth and pose estimation, point tracking, rendering, simulation, and deep learning frameworks.

Develop and scale world-modeling systems that learn from video and other high-dimensional sensor data. Implement self-supervised learning, design architectures for predicting dynamics, build preprocessing pipelines, and create efficient model-based planning and distributed training infrastructure.

Research and develop world-model based AI: self-supervised learning from video, predictive architectures, scalable data curation, evaluation benchmarks, and model-based planning algorithms. Design, run, and analyze experiments in large-scale accelerator environments.