在 UC Berkeley 获博士（导师 Pieter Abbeel 和 Stefano Ermon）。创造了去噪扩散概率模型（DDPM，2020），复兴并极大改进了基于扩散的生成建模方法。DDPM 成为 Stable Diffusion、DALL-E 2、Midjourney、Sora 以及几乎所有现代图像和视频生成系统的基础技术。加入 Google Brain（现 Google DeepMind），继续推进扩散模型及其在视频生成中的应用。
Earned his PhD from UC Berkeley under Pieter Abbeel and Stefano Ermon. Created Denoising Diffusion Probabilistic Models (DDPM, 2020), which revived and dramatically improved the diffusion-based approach to generative modeling. DDPM became the core technology behind Stable Diffusion, DALL-E 2, Midjourney, Sora, and essentially all modern image and video generation systems. Joined Google Brain (now Google DeepMind) where he has continued advancing diffusion models and their applications to video generation.