Lei Li

Also known as Li Lei

Lei Li is an associate professor at Carnegie Mellon's Language Technologies Institute whose work spans multilingual NLP, AI security, large language models, and AI for drug discovery and protein design. After a 2011 Carnegie Mellon PhD on mining co-evolving time series, he was a postdoctoral researcher at UC Berkeley and a principal researcher at Baidu's Institute of Deep Learning. In 2016, he became the founding director of ByteDance AI Lab, where he led development of the VolcTrans machine translation system and the Xiaomingbot writing bot. His VOLT vocabulary-learning paper won the ACL 2021 best paper award, and he joined UC Santa Barbara as an assistant professor in August 2021. He moved to Carnegie Mellon's faculty in 2023, where his work includes robust watermarking for AI-generated text.

Now at: Carnegie Mellon University
Badges: Bridge
Links: Website and X

Timeline

2011–2012

A CMU PhD on time series

After a BS in computer science in Shanghai Jiao Tong University's ACM Honored Class, he earns a PhD in computer science at Carnegie Mellon in 2011, advised by Christos Faloutsos. His dissertation on efficient algorithms for mining co-evolving time series is runner-up for the 2012 ACM SIGKDD dissertation award.
2011–2016

Berkeley postdoc and Baidu

He is a postdoctoral researcher in UC Berkeley's EECS department from 2011 to 2014, and he lectures at the 2014 Probabilistic Programming for Advancing Machine Learning summer school in Portland. From 2014 to 2016, he is a principal researcher at Baidu's Institute of Deep Learning in Silicon Valley.
2016–2017

Founding ByteDance AI Lab

He joins ByteDance in 2016 as the founding director of its AI Lab, where he leads a large research and engineering team working on NLP, machine learning and robotics. He also leads development of Xiaomingbot, an AI writing bot that generates sports news on the 2016 Rio Olympics. The work wins a second-class Wu Wen-tsün Award in AI in 2017.
2019–2020

VolcTrans and WMT20 wins

He receives the China Computer Federation's Young Elite Award in 2019 and leads development of VolcTrans, ByteDance's machine translation system, which serves more than a billion users in 100 languages. In November 2020, his team takes first place in five WMT20 translation directions, including Chinese–English and English–Khmer, along with the parallel corpus filtering task for Khmer and Pashto.
2021

ACL best paper for VOLT

With Jingjing Xu, Hao Zhou, Chun Gan and Zaixiang Zheng, he wins the ACL 2021 best paper award for “Vocabulary Learning via Optimal Transport for Neural Machine Translation,” selected from 3,350 submissions. The paper's VOLT algorithm finds good translation vocabularies 100 times faster than traditional approaches, without expensive trial training.
August 2021

Assistant professor at UCSB

In August 2021, he leaves ByteDance AI Lab to join UC Santa Barbara's Computer Science Department as an assistant professor.
2023–2024

Assistant professor at CMU

He joins Carnegie Mellon's Language Technologies Institute as an assistant professor in 2023. At a January 2024 CyLab seminar, he presents attacks that remove watermarks from AI-generated text and images, along with two robust watermarking techniques for text and models.
As of October 2026

Associate professor at CMU

He is an associate professor at Carnegie Mellon's Language Technologies Institute, working on multilingual NLP, AI security, large language models, and AI for drug discovery and protein design, and he teaches courses on large language model systems and generative AI for biomedicine. His CMU team posts the top scores in the IWSLT 2025 simultaneous speech translation competition. In 2026, he helps organize the third Generative and Agentic AI for Biology workshop at ICML 2026.