Arriving to Cornell Tech in January 2027, Pratyush Maini is an assistant professor of computer science at Cornell Tech and the Cornell Ann S. Bowers College of Computing and Information Science.
Maini studies how the data used to train AI systems shapes what they learn, remember, and how reliably they behave. He treats training data as an object of scientific study rather than a fixed input, developing methods to measure how individual data choices affect model capabilities and behavior. His research spans improving the quality and composition of pretraining data, generating synthetic training data, and understanding and mitigating unwanted memorization. More broadly, he seeks to make the development of foundation models more systematic, efficient, and accountable.
Maini’s research on synthetic data and AI safety has informed pretraining practices at leading AI labs, while his benchmarks for machine unlearning have become widely used throughout the field. His work has appeared at leading machine learning conferences, including NeurIPS, ICML, ICLR, ACL, and COLM.
Before joining Cornell Tech, Maini earned a Ph.D. in machine learning from Carnegie Mellon University, where he was advised by Zico Kolter and Zachary Lipton. He was also a founding member of DatologyAI, where he worked on democratizing data curation for the training of foundation models. His honors include an OpenAI Research Award, a NeurIPS award nomination, and multiple best paper awards at major conference workshops.