Ryan Zhiyu An @ UC Merced

I am a Postdoctoral Scholar in the Department of Computer Science and Engineering at University of California, Merced. I received my Ph.D. in Electrical Engineering and Computer Science at UC Merced in 2025.

My goal is to build Autonomously Safe and Responsible AI. When deploying complex AI systems in safety-critical settings, we usually face a trade-off: either rely on human experts to constantly monitor and override risky or harmful outputs, or let the system operate on its own and accept the risks. The first option is costly and doesn't scale, while the second can be unsafe and unpredictable.

I work on closing this gap by developing methods to align, shield, and verify the behavior of complex AI models, so they can operate safely without human supervision. Some of my representative work (*denotes first author):

• *AAAI 2026: MoralReason: Generalizable Moral Decision Alignment For LLM Agents Using Reasoning-Level Reinforcement Learning

• *L4DC 2025: Disentangling Uncertainties by Learning Compressed Data Representation

• *DAC 2024: Go beyond black-box policies: Rethinking the design of learning agent for interpretable and verifiable HVAC control

• *BuildSys 2023: CLUE: Safe Model-Based RL HVAC Control Using Epistemic Uncertainty Estimation

You can find my work in top venues across ML, Design Automation, and Cyber-Physical Systems including ICLR, AAAI, DAC, KDD, L4DC, BuildSys, and IEEE IoT-J. I received Best Paper Award Runner-Up at BuildSys 2023 and Best Paper Award at HICSS 2025.

Email: zan7 AT ucmerced.edu

[Google Scholar]