Mengyuan (Millie) Wu

 Researcher | Engineer | Designer

First Year CS PhD at Columbia University

Human-AI-Interaction (HAI) x Applied ML Behavioral Health

About Millie

My research journey began in computational neuroscience and psychology, under Prof. Kiyohito Iigaya at Columbia’s Zuckerman Institute, where I explored how aesthetic emotions relate to psychiatric patterns. I later worked with Prof. Sam Gershman at Harvard as summer intern, studying reinforcement learning and anhedonia using multimodal health datasets. 

To bring these insights into real-world systems, I joined the XR and BCI groups led by Prof. Steve Feiner and Prof. Paul Sajda, where I conducted multi-visit XR studies and contributed to the development of a large-scale physiological foundation model. 

Currently, I’m a first-year PhD advised by with Prof. Xuhai “Orson” Xu on Human-AI-Interaction, Applied ML for Behavioral Health. 

Looking ahead to the Ph.D. and beyond, my work aims to unify three research threads:

  1. AI Agents for Self-Exploration and Introspection 

  2. Optimizing Behavioral Patterns through the Quantified Self

  3. AI-assisted Creative Manifestation of Self

Publications

Wu, Mengyuan, Fan, Y., Jiang, Z., Feng, R., Dharmavaram, S., Fallon, S., Polowitz, M., Islam, B., Benson, L., Tung, I., Creswell, J. D., & Xu, X. MindfulAgents: Personalizing Mindfulness Meditation via an Expert-Aligned Multi-Agent System. Submitted to the ACM CHI 2026 

Li, Z.*, He, X.*, Wu, Mengyuan, Tong, Z., Wei, H., Yang, B., Sajda, P., Feiner, S. SwEYEpinch: Exploring Intuitive, Efficient Text Entry for Extended Reality via Eye and Hand Tracking. ACM CHI 2026 

Projects

Expert-Aligned Personalized AI Meditation

Multi-Agent System that generates live, personalized meditation under Expert Aligned Framework

CHI 2026 

Emotion-Aware LLM Psychiatry Session Support

Smart glove enabling real-time affective-aware connection between patients and clinicians via multimodal sensor data.

Law of Attraction AI Journalling

Context aware law-of-attraction LLM that helps user manifest their ideal future 
Won HooHack 2024 2nd Place (Intel AI Track)

EEG-Art for Mindfulness

EEG-derived generative art produced during mindfulness practice.

Won 1st Place NeuReality 2024 Hackathon

Teachings

  • Applied Machines Learning, Course Assistant, Columbia University
  • Advanced Optimization, Course Assistant, Columbia University 
  • Corporate Finance, Teaching Assistant, Columbia University

Interested in Collaboration?

Name