profile photo

Juyang Bai

I am a fourth year Ph.D. student in JHU ECE. I am fortunate to be advised by Prof. Laixi Shi! My research focuses on algorithm–system co-design for efficient and trustworthy foundation-model systems.

My current interests center on self-evolving agent systems that learn and adapt through interaction and feedback. Specifically, I study agentic reinforcement learning and in-context learning for continual adaptation in single- and multi-agent systems.

Actively seeking research internship opportunities!

News

  • Jun 2026MAS-PromptBench is released!
  • Jan 2025One paper is accepted to USENIX SEC 2025.
  • Jun 2023One paper is accepted to IROS 2023.
  • Jul 2022One paper is accepted to UbiComp/ISWC 2022.

Selected Publications

MAS-PromptBench: When Does Prompt Optimization Improve Multi-Agent LLM Systems?
Juyang Bai, Laixi Shi
arXiv, 2026
Lifelong Agents: Learning, Aligning, and Evolving@COLM 2026
[Website] [Paper] [Code]
The Map Is Not the Territory: Embedding-Coverage Blacklists for Safe Diffusion Steering overview
The Map Is Not the Territory: Embedding-Coverage Blacklists for Safe Diffusion Steering
Juyang Bai, Tong Zhou, Shaolei Ren, Xiaolin Xu
ECCV, 2026
[Website] [Paper] [Code]
SafeGuide: Adaptive Inference-Time Safety Control for Diffusion Models overview
SafeGuide: Adaptive Inference-Time Safety Control for Diffusion Models
Tong Zhou, Juyang Bai, Xiaolin Xu, Shaolei Ren
Preprint, 2026
Principled Design for Trustworthy AI@ICLR 2026
[Paper] [Code]
SciFig: Towards Automating Scientific Figure Generation overview
SciFig: Towards Automating Scientific Figure Generation
Siyuan Huang*, Yutong Gao*, Juyang Bai*, Yifan Zhou*, Zi Yin*, Xinxin Liu, Rama Chellappa, Chun Pong Lau, Sayan Nag, Cheng Peng, Shraman Pramanick
arXiv, 2026
* Equal contribution
[Paper]
Phantom: Privacy-Preserving Deep Neural Network Model Obfuscation in Heterogeneous TEE and GPU System
Juyang Bai, Md Hafizul Islam Chowdhuryy, Jingtao Li, Yao Fan, Chaitali Chakrabarti, Deliang Fan
USENIX Security, 2025
[Website] [Paper] [Code]
Learning Representation for Anomaly Detection of Vehicle Trajectories
Ruochen Jiao, Juyang Bai, Xiangguo Liu, Takami Sato, Xiaowei Yuan, Qi Alfred Chen, Qi Zhu
IROS, 2023
[Website] [Paper]
Towards a Toolkit for Free Living Wearable Development
Blaine Rothrock, Alexander Curtiss, Juyang Bai, Josiah Hester
UbiComp/ISWC, 2022
[Website] [Paper]

Misc Projects


Template from Jon Barron. Thanks Jon!