Iโm a Senior Research Scientist and Team Manager at Shopee (Singapore), where I lead a machine learning engineering team dedicated to optimizing merchandise ranking in large-scale e-commerce systems. I am the creator of NSG and SSG algorithm, which is widely used in industry for large-scale vector database.
We are actively seeking collaborations on cutting-edge topics in efficient training and scaling of LLMs, and large transformer-style recommender models (LTRM). We are especially interested in:
- Scaling laws for LTRM systems
- Agent-style modeling for recommender systems
- Integrations of foundation models into personalized product search and ranking
If you are exploring similar directions or have exciting ideas, feel free to reach out!
My research focuses on building Scalable and Practical AI Systems to expand the boundaries of intelligent products in real-world applications. Our interests span multiple dimensions:
Foundation models define the upper bound of AI capabilities. To accelerate deployment and productization, we work on:
- Efficient pre/post-training strategies
- Life-long learning & continual adaptation
- Model compression & distillation
- Beyond-transformer paradigms
- General-purpose ML efficiency
Modern AI agents depend heavily on effective interaction with key tools. We aim to enhance:
- Search engines, recommender systems, and vector databases
- Retrieval-augmented generation (RAG)
- Tool-use optimization for agent systems
Inspired by prior advances in agent collaboration and reinforcement learning, weโre exploring:
- Personalization-aware learning for agents
- Scalable multi-agent systems
- Collaborative decision-making and planning
Our long-term goal is to bridge state-of-the-art research with real-world AI products, improving both efficiency and effectiveness at scale. We believe the future of AI lies in the synergy between powerful foundation models, intelligent tools, and adaptive agent systems.
-
๐ Academic Training
I received both my Ph.D. and Bachelor's degrees from Zhejiang University (ZJU), where I was fortunate to be mentored by Professor Xiaofei He (National Distinguished Young Scholar, former Dean of Didi Research Institute) and Professor Deng Cai (National Excellent Young Scholar).
I also spent time as a Visiting Scholar at the University of Southern California (USC), collaborating with Professor Xiang Ren on research in machine learning and knowledge representation. -
๐ผ Industry Experience
Previously, I worked as an Expert Machine Learning Engineer at Alibaba Group, where I contributed to large-scale AI systems and recommendation technologies powering Alibaba's core platforms.
๐ Academic Profile
Check out my Google Scholar for a full list of publications.
๐ง Open-Source Contributions
Large-Scale Vector Databases
- NSG: Efficient vector retrieval in Euclidean space.
- SSG: Optimized structure for scalable search in Euclidean space.
- Efanna: Fast approximate nearest neighbor search in Euclidean space.
- PSP: Indexing for inner product similarity search.
- MAG: Unified indexing for both Euclidean and inner product similarity.
Recommender Systems
- ResFlow: Low-cost joint learning framework for multi-behavior recommendation.
๐ Books
- Business Driven Recommender Systems: Methodology and Practice ใไธๅก้ฉฑๅจ็ๆจ่็ณป็ป๏ผๆนๆณไธๅฎ่ทตใ
- ๐ง Email: fc731097343 [at] gmail.com
- ๐ผ LinkedIn
- ๐ฌ X
- ๐ฌ Zhihu (็ฅไน)