Jiajun Zhu 「N.B」 My Chinese name is "祝家俊".
It’s also fine to call me Jun.

I am currently a PhD student at University of Texas at Austin, advised by Atlas Wang and Aditya Akella and work closely with Pan Li at Gatech. I previously graduated from Zhejiang University with a major in Mathematics and a minor in Entrepreneurship. I have been working on explainable AI and large language model architectures. Broadly, my research scope is around trustworthiness (e.g., persona steering, preference alignment) and efficiency (e.g., inference acceleration, efficient architectures). I advocate for rigorous evaluation frameworks, grounded in FAIR principles, for next-generation infrastructure (models, data, algorithms, and systems optimizations), as exemplified by Marin. Feel free to reach out if you’d like to collaborate or chat.

Email: jiajunzhu [at] utexas [dot] edu

CV  /  Google Scholar  /  Github

「News」
「Research」
TAPE
Rethinking Addressing in Language Models via Contextualized Equivariant Positional Encoding
Jiajun Zhu*, Peihao Wang*, Ruisi Cai, Jason D. Lee, Pan Li, Zhangyang Wang
ICML 2025.
Paper Code
benchmark
Towards Unveiling Sensitive and Decisive Patterns in Explainable AI with a Case Study in Geometric Deep Learning
Jiajun Zhu, Siqi Miao, Rex Ying, Pan Li
Nature Machine Intelligence.
Paper Code
「Extra」

In addition to my academic interest in AI, I am fascinated by other cutting-edge technologies such as brain-computer interfaces and blockchains. I also enjoy reading biographies, learning from people I admire—especially entrepreneurs—and thinking about how to contribute meaningfully to society. My role models include Charlie Munger for his wisdom, Stephen Curry for his grit, Li Jian (李健) for his serenity, as well as many of my kind, thoughtful, and intelligent peers.

Website template: Jon Barron's and Yunzhi Zhang's