Hui Sun (孙辉)
Ph.D. Student at LAMDA Group, Nanjing University
Supervised by Prof. Ming Li (黎铭)
Biography
I am a Ph.D. student at School of Artificial Intelligence in Nanjing University , and a member of LAMDA Group , led by Prof. Zhi-Hua Zhou .
I received my B.Sc. from Software College at Jilin University in 2019. I then joined Nanjing University as a M.Sc. student (direct admission) and graduated in 2022. From 2022 to 2023, I worked as an algorithm engineer at Shopee . In 2023, I returned to LAMDA to pursue a full-time Ph.D.
Research Interests
I am interested in machine learning and data mining, especially the following:
Publications
Preprint
Loading...
, Zheng Xie, Hao-Yuan He, and Ming Li
Loading...
Xin-Ye Li, Yali Du, , and Ming Li
Journal Papers
Post-Incorporating Code Structural Knowledge into Pretrained Models via ICL for Code Translation
Yali Du†, , and Ming Li
IEEE Transactions on Software Engineering, 2025
Learning Personalizable Clustered Embedding for Recommender Systems
Yizhou Chen, Guangda Huzhang, Anxiang Zeng, Qingtao Yu, , Heng-Yi Li, Jingyi Li, Yabo Ni, Han Yu, and Zhiming Zhou
ACM Transactions on Recommender Systems, 2024
Enhancing Unsupervised Domain Adaptation by Exploiting the Conceptual Consistency of Multiple Self-supervised Tasks
, and Ming Li
SCIENCE CHINA Information Sciences, 2023, 66: 142101
Conference Papers
Dynamic-Static Synergistic Selection Method for Candidate Code Solutions with Generated Test Cases
Ren-Biao Liu, Jiang-Tian Xue, Chao-Zeng Ma, , Xin-Ye Li, and Ming Li
The 40th AAAI Conference on Artificial Intelligence, 2026
ARBench: Algorithmic Reasoner or API Alchemist? Evaluating Code-Generating LLMs beyond API Calls
Ren-Biao Liu, Chao-Zeng Ma, An-Qi Li, , Xin-Ye Li, and Ming Li
The 40th AAAI Conference on Artificial Intelligence, 2026
Ovis2.5 Technical Report
Ovis Team (Contributor: ), Alibaba Group
Preprint, arXiv: 2508.11737
MDP3: A Training-free Approach for List-wise Frame Selection in Video-LLMs
, Shiyin Lu, Huanyu Wang, Qing-Guo Chen, Zhao Xu, Weihua Luo, Kaifu Zhang, Ming Li
IEEE/CVF International Conference on Computer Vision, 2025
Revisiting Chain-of-Thought in Code Generation: Do Language Models Need to Learn Reasoning before Coding?
Ren-Biao Liu, An-Qi Li, Chao-Ding Yang, , and Ming Li
The 42nd International Conference on Machine Learning, 2025
A Joint Learning Model with Variational Interaction for Multilingual Program Translation
Yali Du, , and Ming Li
IEEE/ACM 46th International Conference on Automated Software Engineering, 2024
Ambiguity-Aware Abductive Learning
Hao-Yuan He, , Zheng Xie, and Ming Li
The 41st International Conference on Machine Learning, 2024
Cooperative and Adversarial Learning: Co-Enhancing Discriminability and Transferability in Domain Adaptation
, Zheng Xie, Xin-Ye Li, and Ming Li
The 37th AAAI Conference on Artificial Intelligence, 2023
Semi-Supervised Learning with Support Isolation by Small-Paced Self-Training
Zheng Xie, , and Ming Li
The 37th AAAI Conference on Artificial Intelligence, 2023
Clustered Embedding Learning for Recommender Systems
Yizhou Chen, Guangda Huzhang, Anxiang Zeng, Qingtao Yu, , Heng-Yi Li, Jingyi Li, Yabo Ni, Han Yu, and Zhiming Zhou
The World Wide Web Conference, 2023
Patents
一种面向环境变化的无监督迁移学习图像分类算法
Ming Li (黎铭), , Zhi-Hua Zhou (周志华)
Patent No. 202210461879.4
Awards & Honors
Value Star Awards
Top 4% (8 out of 200+)
Shopee · Dec 2022
Artificial Intelligence Scholarship
50 recipients university-wide per year
Nanjing University · Oct 2019
China Collegiate Computing Contest (CCCC)
Outstanding Winner (Highest Honor)
Jilin Province · Mar 2018
ACM-ICPC Asia Regional Contest
Silver Medal
Nanning · Dec 2017
ACM-ICPC Asia EC-Final
Bronze Medal
Shanghai · Dec 2017
Jilin Province Collegiate Programming Contest
Gold Medal
Changchun · May 2017
Northeast Collegiate Programming Contest
Gold Medal
Changchun · Sep 2016
Work Experience
Alibaba AIDC - Ovis Team
LLM Researcher | Multimodal LLM Research
As a key developer, I contributed to Ovis V1.6, V2.0, and V2.5. Primarily responsible for developing video understanding capabilities from scratch (including data preparation, algorithm design, and model training), contributing to the SFT for the Ovis GUI Agent with a focus on optimizing GUI & general grounding, and collecting trajectory data for the agent's web-based operations.
Ovis 2.5
<40B params
Ovis 2.0
OpenCompass
MDP3
First-author
Ovis2.5
Contributor
Shopee (Sea Limited)
Algorithm Engineer (Full-time) | Ranking for E-commerce Search
Primarily responsible for maintaining and optimizing the foundational E-commerce search ranking model, initially used in fine-ranking and extended to recall, coarse-ranking, and long-tail scenarios. Key work includes:
CEL - CTR AUC
Embedding Learning
PLE - CTR AUC
Multi-task
AutoDis - CTR
Numeric feature
AutoDis - CVR
Numeric feature
Data to Convergence
Module-wise Model Resue
Training Speed
37→76 samples/(cpu·s)
Inference Speed
95.5ms→28.8ms
CEL Speed
Real-world impl.
AliExpress (Alibaba Group)
Algorithm Engineer (Intern) | Ranking for E-commerce Recommendation
Designed a country-specific multitask fine-ranking model for multinational e-commerce platform. During the internship, reproduced 14 papers and finally adopted PLE, ESMM, and Gradient Block to optimize the multitask framework. Designed country-level PLE in the top network to capture specific information of the top 5 key countries, and used DCN-v2 in the bottom network to achieve higher-order feature crossover for country-differentiated features.
Reproduced Papers
CTR AUC
CTR GAUC
L2P AUC
L2P GAUC
ByteDance
Search Algorithm Engineer (Intern) | Toutiao Series Apps
Mainly responsible for the official user search in ByteDance's "Toutiao" series apps (Toutiao, Xigua Video, DongCheDi, etc.), including the vertical search results under the user search tab, and the user card result under the all web search and video search tab.
CTR of query
15%→45%
Card Recall Ratio
User cards
Top 3 CTR
User cards
Education
Ph.D. in Computer Science and Technology, Schoolar of Artificial Intelligence
LAMDA Group · Supervisor: Prof. Ming Li (黎铭)
M.Sc. in Computer Science and Technology, Schoolar of Artificial Intelligence
Recommended admission (without entrance exam); Ranked 1st in coding test during interview.
B.Sc. in Software Engineering (Excellent Engineer Program)
GPA: 3.7/4.0 (Top 5%) · Exchange: ITMO University, Russia (2017)
Teaching Assistant
• Artificial Intelligence Guidance (With Prof. Ming Li; Fall 2025)
• Data Mining for Complex Data Objects (With Prof. Ming Li; Fall 2021)
• Introduction to Data Mining (With Prof. Ming Li; Spring 2020)