I am a Software Engineering undergraduate at Shanghai Jiao Tong University (SJTU), advised by Prof. Tao Huang, and expect to graduate in 2027.
My research experience includes an internship with Prof. Xiaofeng Gao at SJTU on cloud workload prediction, and an internship with Zhuoran Lu at The University of Hong Kong (HKU) on AI-assisted storytelling. In summer 2026, I was a visiting student at the Hong Kong University of Science and Technology (HKUST), supervised by Prof. Xiaojuan Ma.
My interests lie in efficient world models, human–computer interaction, and data mining. I am interested in making AI both more computationally efficient and more useful to people—from efficient video generation to interactive systems that support creativity and communication.
Ultimately, I hope to develop AI technologies that serve people and make their benefits more widely accessible.
IEEE Transactions on Services Computing, 19(3): 2506–2519, 2026.
STDPS models cross-service dependencies, service state, and forecast uncertainty to support more accurate, resource-efficient cloud scaling.
NeurIPS 2026 Creative AI Track · Accepted
FANVERSE turns canon and fan fiction into a navigable branching story universe, helping readers compare story paths and create grounded new branches.
ICLR 2027 submission · Under review
LinMorph learns to transform pretrained attention weights into linear-attention modules, enabling faster recovery and efficient video generation.
Research Intern · Mentor: Zhuoran Lu
Designed and built an LLM-powered storytelling prototype for exploring canon and fan fiction, comparing story states, and creating grounded “What-If” branches. This work led to FANVERSE, accepted to the NeurIPS 2026 Creative AI Track.
Research Intern · Mentor: Prof. Xiaojuan Ma
HCI for Sports. Developing an AI-mediated interactive system to improve communication between coaches and athletes. My work includes requirements analysis, interaction design, and full-stack prototyping. Planned submission to IMWUT in 2027.
Research Intern · Mentor: Prof. Tao Huang
Efficient World Models. Studying linear attention for efficient video generation, with the aim of reducing computational and memory costs. This work led to LinMorph, which learns to transform pretrained attention weights into linear-attention modules. Submitted to ICLR 2027 · Under review.
Research Intern · Mentor: Prof. Xiaofeng Gao
Studied time-series prediction for cloud resource scheduling, including system state, spatio-temporal dependencies, and forecast uncertainty. Contributed to the evaluation of STDPS, published in IEEE Transactions on Services Computing.
School of Computer Science
B.S. in Software Engineering
Shanghai, ChinaAdvisor: Prof. Tao Huang
Department of Computer Science and Engineering
Summer Visiting Student
Hong Kong SAR, ChinaSupervisor: Prof. Xiaojuan Ma
High School
Chengdu, Sichuan, ChinaOutside research, I enjoy badminton, good food, travel, and visits to museums and aquariums. Social practice has also taken me to communities and places across China.
(山海踏歌行 · 揽鹤山风物，谱振兴新篇)
I co-organized an SJTU social practice project in Heshan, Guangdong, exploring local heritage and rural revitalization.
Project award: Grand Award (university level) · My role: One of the organizers
Read the article ↗
(红土新程，壮乡共兴)
I co-organized field research in Guangxi on border-region development and local communities.
Project award: Grand Award (university level) · My role: One of the organizers
Read the article ↗
(沽上百年工，津朝新气象)
Our team explored Tianjin's industrial heritage and waterfront culture through field visits.
Project award: Third Award (university level)
Read the article ↗A few snapshots from my travels and everyday life.
No photographs yet.