Minbiao Han

About

I am a Data Scientist at Microsoft, working on user understanding and personalization to make advertising and recommendation systems more effective and better aligned with user intent. I am part of the Microsoft AI organization, where I build LLM-based profile generation and interest modeling systems, applying techniques such as prompt design, post-training, and evaluation methodology to large-scale behavioral signals.

Prior to Microsoft, I completed my PhD in Computer Science at The University of Chicago, where my research focused on machine learning in strategic, multi-agent settings: designing online learning algorithms that adapt to how agents respond to a model’s decisions, from learning optimal policies through repeated interaction under incentive constraints to learning to value data and model performance in ML marketplaces. That perspective still shapes how I think about modeling user behavior today, learning from feedback that is noisy, adaptive, and shaped by the system itself.