Hi there 
Welcome to my Homepage!
I am Zixuan Weng, a Ph.D. student in Computer Science at the Georgia Institute of Technology, advised by Prof. Ling Liu. I received my B.Eng. in Computer Science and Technology from Beijing Jiaotong University in 2024.
My research interests include large language model safety, agent safety, and agentic reinforcement learning. Feel free to contact me about research collaboration and academic opportunities.
News
- 2026.04 FineSteer was accepted to ACL 2026 as an oral paper (Top 4%).
- 2025.08 Foot-In-The-Door was accepted to EMNLP 2025 as an oral paper (Top 4%).
- 2025.08 Recognized as a KDD Outstanding Reviewer (Top 10%).
- 2025.06 JailbreakDiffBench was accepted to ICCV 2025.
- 2025.05 SDE was accepted to KDD 2025.
- 2024.02 Our work on PPR-based embeddings was accepted to WWW 2024 as an oral paper (Top 9%).
Education
Aug. 2026 - Present
Ph.D. in Computer Science
Advisor: Prof. Ling Liu
Publications
* Equal contribution.
Zixuan Weng*, Jinghuai Zhang*, Kunlin Cai, Ying Li, Peiran Wang, Yuan Tian.
A unified framework for constructing fine-grained, query-specific steering vectors for large language models.
ACL 2026 Oral (Top 4%) [paper]
Zixuan Weng*, Xiaolong Jin*, Jinyuan Jia, Xiangyu Zhang.
A multi-turn jailbreak strategy motivated by the Foot-In-The-Door phenomenon.
EMNLP 2025 Oral (Top 4%) [paper]
Zixuan Weng, Jindong Han, Wenzhao Jiang, Hao Liu.
A simplified state space model with disentangled dependency encoding for time series forecasting.
KDD 2025 [paper]
- ACL 2026 FineSteer: A Unified Framework for Fine-grained Inference-time Steering in Large Language Models.
Oral, Top 4%. [paper] - EMNLP 2025 Foot-In-The-Door: A Multi-turn Jailbreak for LLMs.
Oral, Top 4%. [paper] - KDD 2025 SDE: A Simplified and Disentangled Dependency Encoding Framework for State Space Models in Time Series Forecasting.
[paper] - ICCV 2025 JailbreakDiffBench: A Comprehensive Benchmark for Jailbreaking Diffusion Models.
[paper] - WWW 2024 Towards Deeper Understanding of PPR-based Embedding Approaches: A Topological Perspective.
Oral, Top 9%. [paper] - UbiComp 2023 I Know Your Intent: Graph-Enhanced Intent-Aware User Device Interaction Prediction via Contrastive Learning.
[paper] - Preprint DAMBench: A Multi-Modal Benchmark for Deep Learning-based Atmospheric Data Assimilation.
In submission.
Experience
LLM Jailbreak
Research Assistant · Penn State & Purdue University
Advisors: Prof. Jinyuan Jia and Prof. Xiangyu Zhang
Time Series Forecasting
Research Assistant · Hong Kong University of Science and Technology
Advisor: Prof. Hao Liu
Understanding PPR-Based Graph Embeddings
Research Assistant · Chinese University of Hong Kong
Advisor: Prof. Sibo Wang
User Behavior Prediction with Graph Contrastive Learning
Research Assistant · Tsinghua University
Advisor: Prof. Yong Jiang
Awards
- 2025, KDD Outstanding Reviewer (Top 10%).
- 2023, National College Students’ Innovative Entrepreneurial Training Plan Program (Top 5%).
- 2023, National Scholarship (Top 1%).
- 2023, First Class Scholarship for Academic Performance (Top 2%).
- 2023, Second Class Scholarship for Social Work Performance (Top 5%).
Academic Services
- Session Chair: KDD 2025.
- Reviewer: NeurIPS 2026; CVPR 2026; ARR May 2025; WWW 2026; KDD 2024, 2025, and 2026.
- Volunteer: KDD 2025; EMNLP 2025.