Kevin Zeng Hu

We should all be able to ask questions of our data. Towards this end, I build data visualization and analysis tools that are accessible to the many, not just the few. My approach is to combine human-centered interface design, recommender systems, and machine learning — though the "how?" is secondary to the "why?"

I'm currently a doctoral student with the Collective Learning Group at the MIT Media Lab. Previously, I studied Physics at MIT and interned at several tech companies. My past projects have been published in top academic journals and covered in diverse media outlets. But I'm proudest of having reached the front page of Reddit.

Contact

kevinzenghu@gmail.comTwitterLinkedIn

Projects

Sherlock (2019)
Deep learning for semantic type detection
Info: KDD Preprint
VizML (2019)
Machine learning for visualization recommendation
Info: Blog Post, Video, CHI Paper
VizNet (2019)
Towards a large-scale visualization learning and benchmaring repository
Info: Video, CHI Paper
DIVE (2018)
Recommendation-driven data exploration system
Info: Video, SIGMOD Workshop Paper
Clinton Circle (2016)
WikiLeaks e-mails visualized as network of connections
Press: Univision
Global Language Network (2014)
Understanding the connections between languages
Info: Video, PNAS Paper
Press: The Economist, Science Magazine, Harvard Business Review
Pantheon (2014)
Quantifying historical cultural production
Info: Video, Nature Scientific Data Paper
Press: New York Times, Fast Company, BostInno
GIFGIF (2014)
Measuring emotional content of animated GIFs
Press: Wired, Washington Post, Buzzfeed
FOLD (2014)
Platform for reading and writing non-linear and contextual stories
Press: Nieman Journalism Lab

Education

Last Updated on May 2019