Yash Thube
I care about making machines intuitively understand space, time, physics and adapt to our physical world. Broadly, my interests revolve around 3D Computer Vision, Reinforcement Learning and Embodied/Physical AI.Works shaping my thinking: JEPA/V-JEPA (predictive architectures), NVIDIA Cosmos (world foundation models), DeepMind’s Dreamer & Genie series (imagination-based), Sakana AI’s work in open-endedness, World Labs, Runway (spatial intelligence), Physical Intelligence, Sklid, 1X, Genesis (robot foundation models) and many more!
Currently working on vision at Future Sportler. Always open to collaborations!