Research Scientist, Google DeepMind
2 papers at NeurIPS 2025
Video generalization models can learn physics-based controls from synthetic data and generalize this control across diverse geometries, settings, and materials.
A unified online framework for open-world 3D object extraction.