Rui Wang

Also known as 王瑞

Rui Wang is a natural language processing researcher at Shanghai Jiao Tong University who has worked on machine translation for over ten years and now works on large language models. Born in Harbin, he earned a bachelor's degree from Harbin Institute of Technology in 2009 and pursued a joint doctoral program between Shanghai Jiao Tong University and CNRS in Lyon from 2012 to 2016. From 2016 to 2020, he was a researcher at Japan's NICT, where he coauthored an unsupervised system that ranked first for German-to-Czech translation at WMT 2019. After joining Shanghai Jiao Tong University in 2021, he coauthored the MAPS translation framework (TACL 2024), a study of overthinking in o1-like models (ICML 2025) and the DeepMath-103K dataset (ICLR 2026). He is a tenure-track associate professor in the Institute of Artificial General Intelligence and a senior area chair for NeurIPS 2026.

Now at: Shanghai Jiao Tong University

Timeline

2009–2012

Harbin Institute of Technology degree

Born in Harbin in March 1985, he earns a bachelor's degree from Harbin Institute of Technology in 2009. From 2009 to 2012, he studies and works at the Xinjiang Branch of the Chinese Academy of Sciences.
2012–2016

Joint PhD with CNRS in Lyon

From 2012 to 2016, he pursues a joint PhD program between Shanghai Jiao Tong University and the French National Centre for Scientific Research (CNRS) in Lyon, France.
2013

Converting continuous-space language models

He begins researching language models in 2013, when the field shifts from n-gram approaches to neural networks. The EMNLP 2013 paper 'Converting Continuous-Space Language Models into N-Gram Language Models for Statistical Machine Translation' lists him as first author.
2016–2020

Researcher at NICT in Kyoto

From 2016 to 2020, he is a researcher at Japan's National Institute of Information and Communications Technology (NICT) in Kyoto. NICT's unsupervised system, which he coauthors, ranks first for German-to-Czech translation at WMT 2019 using only the organizers' data. At WMT 2020, the joint SJTU-NICT team's primary systems win first place on English to Chinese, Polish to English and German to Upper Sorbian.
2021

Joining Shanghai Jiao Tong University

In 2021, he joins Shanghai Jiao Tong University. At EMNLP 2021, the tutorial 'Syntax in End-to-End Natural Language Processing' by Hai Zhao, Rui Wang and Kehai Chen surveys syntactic parsing and the role of syntax in semantic role labeling and machine translation.
2024

The MAPS translation framework

He coauthors 'Exploring Human-Like Translation Strategy with Large Language Models', published in TACL in 2024, which proposes the MAPS framework (Multi-Aspect Prompting and Selection). MAPS has a language model first induce keywords, topics and relevant demonstrations for the source sentence, then filters that knowledge with quality estimation before translating.
2025–2026

Overthinking study and DeepMath-103K

In 2025, he coauthors 'Do NOT Think That Much for 2+3=?', which finds that o1-like models spend excessive computation on simple problems and proposes self-training strategies to reduce it; the paper appears at ICML 2025. He also coauthors DeepMath-103K, a mathematical dataset with verifiable answers for reinforcement learning, published at ICLR 2026.
As of October 2026

Associate professor at Shanghai Jiao Tong

He is a tenure-track associate professor in the Institute of Artificial General Intelligence at Shanghai Jiao Tong University, where his Language Intelligence Lab (previously the MT Lab) works on language modeling, computational linguistics and machine translation. He has worked on machine translation for over ten years. In 2026, he serves as senior area chair for NeurIPS, ACL, EACL and EMNLP and as an action editor of TMLR.