Aditi Raghunathan

Aditi Raghunathan is a machine-learning researcher whose work spans certified robustness and reliable AI. She earned a BTech in computer science at IIT Madras in 2016 and completed a Stanford PhD under Percy Liang in 2021, then joined Carnegie Mellon after a Berkeley AI Research postdoc in 2022. Her 2018 paper proposed a semidefinite certificate for neural-network robustness, and her 2025 coauthored study of language-model creativity won ICML's Outstanding Paper Award; she received a Sloan Research Fellowship in 2026. As of October 2026, she is an assistant professor at CMU with affiliations in Computer Science, the Machine Learning Department and the Language Technologies Institute; she leads the AI Reliability Lab and teaches course 15889.

Now at: Carnegie Mellon University
Links: Website

Timeline

2016

Computer science at IIT Madras

She earns a BTech in Computer Science at IIT Madras in 2016.
2018

Certified adversarial defenses

With Jacob Steinhardt and Percy Liang, she proposes a semidefinite relaxation that certifies neural-network robustness in a 2018 ICLR paper.
2021

A Stanford PhD

She completes a PhD at Stanford in 2021 under adviser Percy Liang.
2022

Arthur Samuel thesis award

In 2022, she wins Stanford's Arthur Samuel PhD Best Thesis Award for Adversarially Robust Machine Learning with Guarantees, advised by Percy Liang.
2022

Joining Carnegie Mellon

After a postdoc at Berkeley AI Research, she joins Carnegie Mellon in Fall 2022 as an assistant professor in the Computer Science Department, with an affiliation in the Machine Learning Department.
2025

Language-model creativity research

In 2025, she coauthors “Roll the dice & look before you leap,” which uses minimal algorithmic tasks to study open-ended creativity in language models. The paper argues that teacherless training and diffusion models produce more diverse and original output than next-token learning, and it wins ICML’s Outstanding Paper Award.
2026

A Sloan Research Fellowship

In 2026, she receives a Sloan Research Fellowship as one of five Carnegie Mellon University recipients.
As of October 2026

Assistant professor at CMU

She is an assistant professor in CMU's Computer Science Department, with affiliations in the Language Technologies Institute and Machine Learning Department. She leads CMU's AI Reliability Lab, which builds reliable, aligned and trustworthy AI through rigorous analysis and principled methods; she teaches course 15889 in Fall 2026.