Roger Zhe Li

I am a Senior Research Scientist at Huawei Ireland Research Center, where I also serve as a tech lead for recommendation and advertising research. My work sits where information retrieval meets large-scale industrial systems. Lately that means two things. One is semantic IDs: how to construct item identifiers that carry meaning, and how to consume them as features in retrieval-based ranking systems. The other is privacy-aware ad recommendation: keeping models accurate when personalized features are limited or must stay on the device.

Before Huawei, I spent several years as a Senior Data Scientist and Product Owner at adidas in Amsterdam. There I led computer-vision work on product representations and generative AI, and built recommendation and supply-chain solutions. I received my PhD in Computer Science from TU Delft, supervised by Julián Urbano and Alan Hanjalic. My thesis was on metric optimization and mainstream bias in recommender systems. Before that, I studied at Tianjin University (M.Sc., cum laude) and Zhejiang University (B.Eng.).

Research interests

Recommender systems

• Semantic ID construction and item tokenization
• Semantic ID features in retrieval and ranking
• Cross-domain, multimodal and cold-start recommendation

Computational advertising

• Ranking and CTR/CVR prediction under privacy and feature constraints
• Cloud–device collaborative modeling

Evaluation and fairness

• Evaluation methodology and metrics for recommender systems
• Fairness and bias in recommendation