Bio

I am a research scientist at Visa Research. As a member of the Foundational AI team, I am currently working on payment foundation models, agentic fraud detection, and risk modeling for agentic commerce. My research interests include transaction data modeling, ML trust and safety, and graph mining.

Prior to joining Visa, I obtained my Ph.D. in Computer Science at University of Illinois Chicago in 2022, advised by Prof. Philip S. Yu. My Ph.D. dissertation is “Robust Graph Learning for Misbehavior Detection” [PDF]. Before attending UIC, I received my B.E. from Beijing University of Posts and Telecommunications in 2017.

News

  • [11/2026] Release the VRLLM technical report on turning a 1B LLM into a powerful tabular classifier [Website].
  • [11/2026] One agentic anomaly detection paper PersonaAgent is accepted by KDD 2027.
  • [09/2026] One LLM-based feature engineering paper is accepted by IJCNLP-AACL 2026 Findings.
  • [05/2026] The conference version of TransactionGPT is accepted by KDD 2026.
  • [05/2026] Serving as a PC member for Workshop on AI for Fraud and Abuse at KDD 2026.
  • [04/2026] One KV-cache compression paper FAFO is accepted by ICML 2026.

Show more

Talks and Blogs

Show more

Recent Work

  • VRLLM: Turning LLMs into Powerful Tabular Classifiers Tech Report 11/2026
    Yingtong Dou*, Shubham Jain* et al.
    [Paper][Website]
  • PersonaAgent: Retrieval-Augmented Transaction Anomaly Scoring with Refreshable Multi-Entity Persona Memory KDD 2027
    Zhengyu Fang, Liang Wang, Yingtong Dou, Yuzhong Chen et al.
    [Paper]
  • Optimizing Reasoning LLMs for Automated Feature Engineering via Multi-Stage Direct Preference Optimization IJCNLP-AACL 2026
    Mingzhi Hu, Menghai Pan, Yuzhong Chen, Junpeng Wang, Yingtong Dou et al.
    [Paper]
  • TransactionGPT KDD 2026
    Yingtong Dou, Yuzhong Chen et al.
    [Full Technical Report][Conference Version][Slides][Website]
  • FAFO: Lossy KV Cache Compression for Lossless Inference Acceleration via Draftless Fumble Decoding ICML 2026
    Hoang Anh Duy Le, Shaochen Zhong, Yifan Lu, Yingtong Dou et al.
    [Paper][Code]

Selected Publications