About

I’m a year 2 undergraduate student from Hong Kong, studying at Fudan University. My research interests mainly focus on agentic AI and LLM memory.

Education

Fudan University
Bachelor of Science in Software Engineering Sep 2024 – Jun 2028


Publication

* denotes co-first authors

ViDove: A Translation Agent System with Multimodal Context and Memory-Augmented Reasoning
Yichen Lu*, Wei Dai*, Jiaen Liu*, Ching Wing Kwok*, Zongheng Wu*, Xudong Xiao, Ao Sun, Sheng Fu, Jianyuan Zhan, Yian Wang, Takatomo Saito, Sicheng Lai
EMNLP 2025 Demo. [arXiv] [GitHub]

SpeechGPT 2.0-preview
Hanfu Chen, Ke Chen, Qinyuan Cheng, Mingshu Chen, Ruifan Deng, Liwei Fan, Zhaoye Fei, QingHui Gao, Yitian Gong, Ching Wing Kwok, Kexin Huang, Yaozhou Jiang, Xingyu Lu, Shimin Li, Zhengyuan Lin, Ruixiao Li, Qian Tu, Jin Wang, Yang Wang, Siyin Wang, Zhe Xu, Chenchen Yang, Donghua Yu, Yuqian Yao, Yucheng Yuan, Chufan Yu, Dong Zhang, YiWei Zhao, Yuqian Zhang, Jun Zhan, Xin Zhang, Xingjian Zhao, Chengyang Zhu
[Article/Blog] [GitHub] [Demo Page] [HuggingFace]

Scholarship

Hong Kong Young Technology and Innovation Talent(香港青年科創英才)(2024)

The scholarship is awarded to only 100 students each year and aims to provide opportunities for Hong Kong students who have achieved excellent academic results and aspire to engage in sci-tech innovation.

Bao Steel Scholarship(宝钢优秀学生奖学金)(2024-2025)

Each year, 500 students are selected from over 120 universities nationwide with Fudan University allocated only 6 spots in total (2 of which are for students from Taiwan, Hong Kong, and Macau). This fund is one of the most prestigious and nationally recognized educational awards in the country, established with capital from State-Owned Enterprises (SOEs), supported and guided by the government, advised and planned by experts, and actively participated in by universities.

Fudan University Open Source Pioneer Award(复旦大学开源先锋奖)(2025)

Be awarded as the only undergraduate student from College of Computer Science and Artificial Intelligence to reach the finals and win the award, and received the Best Popularity Award (awarded to only one person among all participants, with no other level awards)

An award established by Fudan University to recognize students who have made contributions in the field of open source.

October Excellent Student Award - Honor Track(十月评优优秀学生荣誉创新试点获得者)(2024)

Recognized for outstanding academic performance and innovative contributions during the October evaluation period

Open Source Contributions

Obsidian LiveSync (11,000+ stars)
2nd-ranked contributor and active contributor. Obsidian LiveSync is a popular plugin for real-time synchronization of Obsidian vaults.
[GitHub]

codex-mobile (700+ stars)
26 contributions, 3rd-ranked contributor. A mobile client for OpenAI Codex.
[GitHub]

CodexOnAnyDevices (100+ stars)
Repository owner. A project enabling Codex usage across various devices.
[GitHub]

OpenTreeHole + DanXi-Dev (600+ stars combined)
Served as a core maintainer of OpenTreeHole and DanXi-Dev, whose team maintains a leading campus service application at Fudan University with over 10,000 monthly active users. Contributed to backend development, PR review, server maintenance, and AI agent development.
[OpenTreeHole] [DanXi-Dev] [Website] [App Store] [Play Store]


Project Experience

Danta Treehole
Since 2025, I’ve led the backend department of the Danta Team, the largest Computer Science and development team at Fudan University. Our team develops an internal forum for Fudan students, which boasts an average of 6,000 active students and over 110,000 posts monthly. Beyond leading the backend department, I’ve also engaged in server maintenance, HR management and AI agent design for the team.
[Website] [App Store] [Play Store] [GitHub(backend Repo 1)] [Github(backend repo2)]

World Model Project
Participated in a World Model project: sampled multiple candidate action chunks and their corresponding future trajectories from identical simulator states to train an action-conditioned latent dynamics model; performed autoregressive latent rollouts to predict the long-term consequences of different action sequences; fine-tuned the VLA policy with GRPO using predicted trajectory returns to improve long-horizon planning.

Talks

LLM-based AI Agent - AI Agent 前沿研究與應用探索

video poster
talks_pic

Traditional Reinforcement Learning - 基於星海爭霸 2 的人工智慧 - Google Deepmind 的遊戲強化學習研究

video

Software Development - 復旦大學 FDUCSLG 講座 - 邁向大廠開發崗:學校中的信息差與開發能力

video poster
talks_pic