Jiajun Yu
Jiajun Yu
jiajunyu@zju.edu.cn | +86 13031180113 | Google Scholar | Shanghai, China
Summary
I am a second-year Ph.D. student jointly trained by Zhejiang University and Shanghai Innovation Institute. My research focuses on AI for Science, large language models (LLMs), and graph neural networks (GNNs), with recent work on LLM agents, scientific idea generation, molecular optimization, virtual screening, molecular property prediction, structure prediction, and protein-protein interaction (PPI) prediction. I also have strong large-scale data engineering and model training experience, including building TB-scale datasets from scratch and training and evaluating models on billions of examples.
Professional Experience
2024.08 - 2024.12
Ant Group / Alipay
Algorithm Intern · Shanghai
- Fine-tuned Llama 3.1, Qwen 2.5, GLM-4, and DeepSeek for QA extraction in an employee assistant scenario; the system was shipped to production and led to a SIGIR 2026 paper.
- Designed an entity-extraction-based RAG algorithm that improved multi-task QA extraction performance and led to an ACL 2025 paper.
2023.06 - 2023.12
CUHK-Shenzhen / Shenzhen Institute of Big Data
Research Intern
- Improved readout functions for molecular property prediction from a kernel-method perspective, leading to an IJCAI 2024 paper.
2021.03 - 2021.09
Kuaishou Technology
Algorithm Intern · Beijing
- Led data collection and maintenance for multimodal video retrieval algorithms.
- Built a news short-video generation pipeline from scratch, including 100 million Chinese image pairs and 5 million video-text pairs.
- Participated in the TRECVID 2021 video retrieval track; the team won second place.
Education
2024.09 - 2028.03
Zhejiang University & Shanghai Innovation Institute
Ph.D. in Artificial Intelligence
Advisors: Prof. Haishuai Wang and Associate Prof. Yu Kang.
2021.09 - 2024.06
China Agricultural University
M.Sc. in Computer Science and Technology
Recipient of the National Scholarship twice (top 0.2%).
2017.08 - 2021.05
China Agricultural University
B.Sc. in Computer Science and Technology
Team leader of the university's first MCM/ICM Finalist team in 2020.
Publications
*and+denote co-first author and corresponding author, respectively.
Representative First-Author Work
KDD 2026 Oral, CORE A*
Is Sequence Generation Optimal for Molecular Editing? A Graph-based MoE Alternative.
SIGIR 2026 CORE A*
Code-Based English Models Surprising Performance on Chinese QA Pair Extraction Task.
EMNLP 2026 Main Conference, CORE A*
PRISM: Entropy-Guided Multi-View Exploration for Scientific Idea Generation.
EMNLP 2026 Findings, CORE A*
SkillMol: Skill-Conditioned Action Editing for Molecular Optimization.
IJCAI 2025 CORE A*
A Centrality-based Graph Learning Framework.
IJCAI 2024 CORE A*
Kernel Readout for Graph Neural Networks.
Other First-Author Work
arXiv preprint 2025
Collaborative Expert LLMs Guided Multi-Objective Molecular Optimization.
Neurocomputing Q1, 2024
AGCL: Adaptive Graph Contrastive Learning for Graph Representation Learning.
Knowledge-Based Systems Q1, 2023
MLGAL: Multi-level Label Graph Adaptive Learning for Node Clustering in the Attributed Graph.
The Web Conference CORE A*, 2022
User Donations in Online Social Game Streaming: The Case of Paid Subscription in Twitch.tv.
Knowledge-Based Systems Q1, 2025
Relation-aware Multiplex Heterogeneous Graph Neural Network.
Under review 2026
Retrieval-Augmented Graph Neural Networks for Hard Graph Classification.
Collaborative Publications
Under review 2025
Systematic Discovery of Single-Cell Protein Networks in Cancer with Shusi.
ICML 2025 CORE A*
MetricEmbedding: Accelerate Metric Nearness by Tropical Inner Product.
IJCAI 2025 CORE A*
Divide and Conquer: Coordinating Multiplex Mixture of Graph Learners to Handle Multi-Omics Analysis.
ACM MM 2025 CORE A*
Where Views Meet Curves: Virtual Anchors for Hyperbolic Multi-View Graph Diffusion.
IJCAI 2025 CORE A*
Multi-Omics Analysis for Cancer Subtype Inference via Unrolling Graph Smoothness Priors.
ACL 2025 CORE A*
EXPLAIN: Enhancing Retrieval-Augmented Generation with Entity Summary.
CogSci 2024
Navigating Brain Language Representations: A Comparative Analysis of Neural Language Models and Psychologically Plausible Models.
Engineering Applications of Artificial Intelligence Q1, 2021
Multi-task Support Vector Machine with Pinball Loss.
Honors and Awards
- Young Science and Technology Talent Development Program, Ph.D. Track, CAST, 2025.
- Bronze Award, Stanford RNA 3D Folding Competition 2025 (89/1,489).
- National Scholarship (2022, 2023).
- Outstanding Student Award, China Agricultural University (2022, 2023).
- First-Class Graduate Scholarship (2022, 2023).
- Finalist, MCM/ICM 2020.
- First Prize, Minsheng Cup Mathematical Modeling Competition 2019.
- First Prize, National Undergraduate Mathematical Contest in Modeling, Beijing Division, 2019.
- Honorable Mention, MCM/ICM 2019.
Academic Service
Reviewer for KDD 2025-2026, WWW 2025-2026, AAAI 2026, ICML 2024-2026, IJCAI 2025-2026, ICLR 2025-2026, NeurIPS 2024-2026, ACM MM 2024-2026, Pattern Recognition, Neural Networks, Knowledge-Based Systems, and Neurocomputing.
Visitor Stats: This page views | Site total views | visitors