I am an assistant professor of computer science at UCLA. I lead the Machine Intelligence (MINT) group, where we develop AI systems that can interact and reason with limited supervision. My current research is at the intersection of generative models and sequential decision making. On the applied side, I actively ground this research for developing systems for scientific discovery, particularly in climate and sustainability domains via the ML4Climate initiative.

Previously, I spent a gap year as a research scientist at the Fundamental AI Research (FAIR) Group at Meta. I completed my postdoctoral training at UC Berkeley, PhD at Stanford University, and bachelors at IIT Delhi, all in computer science. During my PhD, I spent wonderful summers interning at Google Brain, Microsoft Research, and OpenAI.

Recently, I co-founded Inception Labs to develop a new generation of parallelizable LLMs that redefine the Pareto frontier for quality, speed, and cost.

recent news

• Our team at Inception Labs launched Mercury, the world’s first commercial-scale diffusion large language models.
• Honored to be included in the Forbes 30 Under 30 List.
• Honored to receive the GOLD Award for Distinguished Young Alumni from IIT Delhi.
• Honored to receive the NSF Career Award.
• We released Stormer, which beats Graphcast to set a new state-of-the-art for ML-driven weather forecasting. Stormer won the best paper award at the Climate Change in AI (CCAI) Workshop at ICLR 2024. Congrats Tung!
• VideoCon won the best paper award at Data Problems for Foundation Models workshop at ICLR 2024. Congrats Hritik!
• Honored to be awarded the Schmidt AI2050 Early Career Fellowship.
• Honored to be selected as a Kavli Fellow by the US National Academy of Sciences.
• Thank you Washington Post, MIT Tech Review for covering our research on ClimaX. My comments on the field more broadly in Nature and Science
• ClimaX wins the best paper award at ICML Workshop on ML and Scientific Modeling. Congrats Tung!
• CleanCLIP wins the best paper award at ICLR Workshop on Trustworthy and Reliable Large-Scale ML. Also to be presented as oral presentation (acceptance rate: 1.8%) at ICCV. Congrats Hritik!