Tatsunori Hashimoto

Timeline

1. 2016

   A doctorate at MIT

   He completes a 2016 PhD thesis at MIT CSAIL, where graduate study is co-advised by Tommi Jaakkola and David Gifford. Undergraduate study in statistics and math at Harvard is with Edoardo Airoldi.

2. 2018

   Fairness without demographics

   With Megha Srivastava, Hongseok Namkoong, and Percy Liang, he publishes Fairness Without Demographics in Repeated Loss Minimization. ICML names the paper a best paper runner-up in 2018. The approach uses distributionally robust optimization to control minority-group risk while remaining oblivious to group identity.

3. 2020

   Joining the Stanford faculty

   Stanford Engineering welcomes him in 2020 as an assistant professor of computer science. A Stanford postdoc with John C. Duchi and Percy Liang, on tradeoffs between average and worst-case performance, precedes the appointment.

4. 2021

   A foundation models report

   The 2021 report On the Opportunities and Risks of Foundation Models lists him among the authors. It describes models trained on broad data at scale and adaptable to many downstream tasks, and it surveys their opportunities and risks.

5. 2022

   A CHI best paper

   He coauthors Jury learning: Integrating dissenting voices into machine learning models with Mitchell Gordon, Michelle Lam, Joon Park, Kayur Patel, Jeff Hancock, and Michael Bernstein, a best paper at CHI in 2022.

6. 2023

   Alpaca and AlpacaFarm

   Alpaca 7B is a model fine-tuned from LLaMA 7B on 52K instruction-following demonstrations, and he is an author. AlpacaFarm, with Yann Dubois, Xuechen Li, Rohan Taori, Percy Liang, and other coauthors, appears at NeurIPS in 2023. He receives a 2023 Amazon Research Award for Alpaca farm, an open framework for language model safety research and development.

7. 2024

   An ICLR honorable mention

   Proving Test Set Contamination in Black-Box Language Models, with Yonatan Oren, Nicole Meister, Niladri Chatterji, and Faisal Ladhak, receives an ICLR 2024 honorable mention. The paper proposes a procedure for detecting test set contamination of language models, with exact false positive guarantees and without access to pretraining data or model weights.

8. As of October 2026

   Courses on language models

   He is an assistant professor of computer science at Stanford and a faculty affiliate of the Institute for Human-Centered Artificial Intelligence. His research uses tools from statistics to make machine learning systems more robust and trustworthy, especially large language models. For 2026–27 he teaches Language Modeling from Scratch, Natural Language Processing with Deep Learning, and Deep Learning Alchemy, and the group builds instruction-following language models, membership inference, and benchmarks for LM safety and capabilities.