I joined Amazon as an Applied Scientist in July 2026. Previously, I worked at Meta from July 2025 to June 2026. I received my Ph.D. in Computer Science in 2026 and my B.S. in Computer Science in 2021.

I am a strong believer in agentic AI. In early 2026, while I was at Meta and experiencing the company's AI-native transformation from top to bottom, I realized that AI had already become intelligent and capable enough to take over a large amount of knowledge work. What is more, its capabilities continue to expand toward increasingly difficult problems and broader domains of human knowledge, while agents can be replicated almost without limit and work in parallel around the clock. Hence, I do not think there is any realistic way for me to compete with that in the long run. A lot of what we build and value today will either become meaningless or be fundamentally redefined. I kept wondering what my own value would be in this new era, and eventually came to a somewhat ironic conclusion: build agentic systems that automate as much of my own work as possible.

My current goal is to build self-evolving agents that can eventually take over everything I used to do manually, both at work and in daily life. To get there, I am working across multiple parallel directions: systems optimization for latency, memory footprint, and resource efficiency; context management through memory and RAG; model optimization through pretraining and post-training; and multi-agent coordination through protocols such as A2A and ACP. At Amazon, I have been applying these ideas to automated detection, localization, and root-cause analysis for online Prime Video services.

Before this shift, I spent several years doing systems research, primarily on emerging real-time communication systems, such as real-time human--GenAI interaction and immersive telepresence. During my Ph.D., I published 27 papers in systems venues, including 15 papers at top conferences as defined by CSRankings. Among them, 16 are first-author papers, including 8 at top conferences. I spent the final year of my Ph.D. at Meta, where I published four papers, including two at top-tier conferences, and contributed to internal product improvements driven by my research efforts. As of September 2026, my publications have received 1,013 citations, including 869 citations to my first-author work. These numbers matter much less to me now, but they still reflect my ability to learn quickly, identify new problems, and turn new ideas into working systems. As publications are no longer a primary focus of my career, this website will be updated less frequently.

Technical Program Committee
ACM ImmerCom

Selected Review Experience
IEEE/ACM Transactions on Networking (ToN), ACM Transactions on Graphics (ToG), IEEE Internet of Things Journal (IoT-J)
ACM UbiComp, IEEE VR

Acknowledgments in My Ph.D. Dissertation

Last Updated: 09/2026