Together AI

Earned a B.Sc. in Machine Intelligence from Peking University (2006–2010) and a PhD in Computer Science from the University of Wisconsin-Madison (2010–2015), both advised by Christopher Ré. Completed a postdoctoral fellowship at Stanford before becoming an Associate Professor in Computer Science at ETH Zurich. Co-founded Together AI in June 2022 as CTO, building the company's technical foundations in distributed systems and data-centric ML. Has instrumental expertise in decentralized and distributed training as well as data-centric ML Ops. Work recognized with SIGMOD Best Paper Award, SIGMOD Research Highlight Award, Google Focused Research Award, and an ERC grant. Employs a systematic, bottom-up research philosophy—an approach that informs Together AI's infrastructure-first strategy for democratizing access to large-scale AI systems.

Associate professor of computer science at Stanford University, focused on research spanning database theory, database systems, and machine learning, with best papers at PODS 2012, SIGMOD 2014, and ICML 2016. MacArthur Fellow (2015). Prior startup experience includes SambaNova and Snorkel, along with Lattice (DeepDive) and Inductiv (HoloClean), acquired by Apple in 2017 and 2020 respectively. Co-founded Together AI in June 2022 alongside co-founders Ce Zhang (CTO) and Percy Liang. Ré's work on data management and ML systems directly informs Together's architecture—he focuses on learning from weak forms of supervision and how ML changes software and hardware systems. Known for being hard-working, intensely focused, humble, and running away from hype and self-promotion.

B.S. from MIT (2004) and Ph.D. from UC Berkeley (2011), followed by a postdoctoral year at Google founding the semantic parsing team. Later served as Lead Scientist at Semantic Machines, a conversational AI startup. Now Associate Professor at Stanford and director of CRFM, where he focuses on making foundation models more accessible through open-source and understandable through rigorous benchmarking. His research background in semantic parsing, robustness, and interpretability directly informed Together AI's mission around reproducibility and open development. Known for wanting to understand things deeply and being a strong proponent of efficient and reproducible research—a philosophy reflected in the company's emphasis on transparency and research-driven infrastructure.

Completed a Ph.D. in Computer Science at Stanford University, where his dissertation was co-advised by Christopher Ré and Stefano Ermon. Joined from Stanford Artificial Intelligence Laboratory (SAIL) where he focused on efficient training of AI models. Created FlashAttention, an open-source library that became a key tool to train foundation models more efficiently—work that received the ICML 2022 Outstanding Paper runner-up award. Now serves as Assistant Professor of Computer Science at Princeton University, leading the Dao AI Lab. FlashAttention has become the de facto standard attention algorithm across all major AI organizations and helped extend LLM context lengths from 4K to hundreds of thousands of tokens. Research focuses on hardware-aware algorithm design at the intersection of machine learning and systems efficiency.

Studied mathematics, physics and computer science at St. Stephen's College, Delhi but dropped out to pursue software design. Founded three exits: created Vipul's Razor, a popular open-source anti-spam system, earning MIT Technology Review's Top 100 young innovators at age 25; co-founded Cloudmark with Napster's Jordan Ritter to commercialize Razor; and co-founded Topsy, a social-media search company acquired by Apple in 2013 for over $200M, where he served as Director of Engineering leading search programs. Applied deep expertise in large-scale data processing at Topsy and Apple to Together AI's infrastructure for efficient AI. Considers himself a cypherpunk with early work in applied cryptography—a founding philosophy reflected in Together's decentralized, open-source orientation.