Anandkumar was born in Mysore. Her parents are both engineers, and her grandfather was a mathematician.[1] Her great-great-grandfather was the Sanskrit scholar R. Shamasastry. She began to study Bharatanatyam and she learnt this style of dancing for many years.[2] She studied electrical engineering at the Indian Institute of Technology Madras and graduated in 2004.[1] She joined Cornell University for her graduate studies, earning a PhD under the supervision of Lang Tong in 2009. Her first project looked at distributed statistical estimation.[3] She was an IBM Fellow at Cornell University between 2008 and 2009. Her thesis considered Scalable Algorithms for Distributed Statistical Inference.[2] During her PhD she worked in the networking group at IBM on end-to-end service-level transactions. She was a postdoctoral scholar at Massachusetts Institute of Technology until 2010, where she worked in the Stochastic Systems Group with Alan Willsky.[4]
Anima Anandkumar has also developed AI algorithms that with applications in various scientific domains including weather forecasting, drug discovery, scientific simulations and engineering design.[21] She invented Neural Operators that extend deep learning to modeling multi-scale processes in these scientific domains and learn in function spaces and are orders of magnitude faster than traditional simulations. She has developed AI-based high-resolution weather models,[22] an AI-aided method for designing anti-infection medical catheters.[23] Neural operators were featured as a highlight for 2021 in Math and Computer Science by the Quanta Magazine,[24] and genome-scale foundation models with emergent behavior in predicting evolutionary dynamics and protein function in several diverse tasks and scenarios,[25] which won the Association for Computing Machinery (ACM) Gordon Bell Special Prize for High Performance Computing-Based COVID-19 Research in 2022.[26]
Anandkumar has also done some of the early work on generalist AI agents using language models, which are capable of life-long learning using foundation models in an interactive manner. In particular, her work has shown how interactive in-context learning in language models can be used to construct actions in form of program code to solve complex open-ended tasks in environments such as Minecraft[27] and robotic reinforcement learning.[28]
While at Caltech, Anandkumar co-founded the AI for Science initiative in 2018. In 2023, she was invited by the Presidential Council of Advisors on Science and Technology (PCAST) on AI+Science.[29] In addition, she has given keynotes at the Annual Meeting of the US National Committee for Theoretical and Applied Mechanics,[30] the UCLA distinguished seminar,[31] the SIAM annual meeting,[32] the Nature Reviews Physics, hosted by the Alan Turing Institute,[33] and the TED2024 conference.[34]
Anandkumar has won several awards and honours, including:[40]
2025 Time100 Impact Award "for using AI to accelerate scientific discovery"[41]
2025 IEEE Kiyo Tomiyasu Award "for contributions to AI, including tensor methods and neural operators with applications to scientific domains"[42]
2024 Blavatnik Award for Young Scientists for "groundbreaking advancements in AI to address practical scientific challenges, drastically accelerating simulation of complex phenomena like weather forecasting, scientific simulations, engineering design and scientific discovery"[43]
2024 TED Speaker on "AI that connects the digital and physical worlds"[44]
2024 Distinguished Alumnus Award by IIT Madras for "for her achievements and contributions towards interdisciplinary scientific innovation"[45]
2023 Guggenheim Fellow in the field of computer science[46]
2023 Schmidt Sciences AI 2050 Senior Fellow that supports established leaders who have made significant contributions to their field[47]
2023 AAAI Fellow for "significant contributions to machine learning including neural operators for scientific machine learning and tensor methods for probabilistic models"[48]
2022 ACM Fellow for "contributions to tensor methods for probabilistic models and neural operators"[49][50]
^Pathak, J., Subramanian, S., Harrington, P. B., Raja, S. S., Chattopadhyay, A., Mardani, M., Kurth, T., Hall, D., Li, Z., Azizzadenesheli, K., Hassanzadeh, P., Kashinath, K., Anandkumar, A. (22 February 2022). "FourCastNet: A Global Data-driven High-resolution Weather Model using Adaptive Fourier Neural Operators". PASC '23: Proceedings of the Platform for Advanced Scientific Computing Conference. abs/2202.11214. arXiv:2202.11214.
