About Me
Ziqi Huang | 黄子棋
Hi there! I am Ziqi Huang, a fourth-year undergraduate student at Tongji University, majoring in Information Security, Guohao Academic. I am a M.S. student Candidate at the Intelligent Information Processing Lab, Department of Computer Science and Technology, Tongji University (Shanghai, China), under the supervision of Prof. Zhihua Wei.
My research interests lie at the intersection of AI safety, AI Deception, Interpretability. Specifically, I’m passionate about:
- SAE and other interpretability methods
- Reinforcement Learning & Alignment
- I’m not very good at CV and large multimodal models. :(
Email: 2253726 [at] tongji.edu.cn / hzq1915851440 [at] gmail.com
News
Publication
- indicates equal contribution
Toward Safer Large Language Models via Internal Mechanism Analysis
Ziqi Huang, Author B, Author C
xxxx 2026
A brief summary: we study interpretable internal representations and propose a practical safety-alignment method for reducing deceptive behavior in LLMs.
𝓜𝓲𝓼𝓬𝓮𝓵𝓵𝓪𝓷𝓮𝓸𝓾𝓼
⚽ 𝓢𝓹𝓸𝓻𝓽𝓼
- Badminton, soccer, and basketball
- Big fan of Stephen Curry and the Golden State Warriors
🎵 𝓐𝓻𝓽𝓼 & 𝓒𝓻𝓮𝓪𝓽𝓲𝓿𝓮 𝓘𝓷𝓽𝓮𝓻𝓮𝓼𝓽𝓼
- Playing the Ukulele and Violin
- Photography, Singing, Cooking