Tri Dao

Tri Dao is an Assistant Professor in the Department of Computer Science at Princeton University. He is also the chief scientist of Together AI. Dao completed his Ph.D. in Computer Science at Stanford University in 2023. His research focuses on machine learning, particularly at the intersection of machine learning and systems. His research interests include sequence models with long-range memory and structured matrices for compact deep learning models. His work includes FlashAttention and Mamba. He is also recognized as an AI2050 Early Career Fellow by Schmidt Sciences.

Tri Dao is a researcher at Princeton University and Together AI, specializing in machine learning and systems. His work focuses on developing efficient and scalable models for sequence modeling, attention mechanisms, and large language models. He has contributed to significant advancements in state space models, such as Mamba and Hippo, and has explored memory-efficient attention mechanisms like FlashAttention. His research also includes the development of frameworks for language model inference, such as Medusa and Starcoder, and the exploration of data augmentation techniques. Dao's work emphasizes the integration of structured state space models with modern deep learning architectures to improve performance and efficiency.