Deeply curious about challenging problems in generative modeling, especially image and video generation, with the goal of offering insights and proposing principled methods that benefit the public. Recently, my work has focused on understanding and advancing unified multimodal LLMs.