Last updated: 31 August 2026

I am a second-year Ph.D. student in Computer Science at Dartmouth College. I am a member of the SAHAS Lab, where I am honored to be advised by Prof. Nikhil Singh and work at the intersection of AI Agents and Vision Science.

I worked as a research intern at Alibaba DAMO Academy, mentored by Dr. Tian Zhou. I also served as a Visiting Student with Prof. Ming-Hsuan Yang at UC Merced and Google DeepMind and Prof. Chao Ma at Shanghai Jiao Tong University. I have previously collaborated with Prof. Lu Qi and Dr. Xiangtai Li at TikTok, as well as with Prof. Junwei Liang at AI Thrust, HKUST(GZ).

My Ph.D. research builds visual intelligence for AI agents. "The limits of my language mean the limits of my world," wrote Wittgenstein — yet a picture is worth a thousand words precisely because images carry a granularity that language has no words for. I study how agents can perceive, remember, and reason at that granularity.

I approach this as a science of measurement for AI agents' vision: diagnose first, then build. My first work, DMV-Bench (EMNLP 2026), makes visual memory measurable by construction: what the agent must later recall never reaches any text channel. That diagnosis motivates DualMem, a dual-coding memory that keeps both a visual and a verbal trace of every observation.

My research interests span:

• (1) Visual memory for agents
• (2) Agentic reinforcement learning
• (3) Long-horizon reasoning and decision-making

I'm open for academic collaborations and actively seeking AI Agent-related summer internship opportunities based in the U.S. for 2027. Please feel free to email me at tangyujin0275@gmail.com or yujin.tang.gr@dartmouth.edu.