email: mao@parsnip.ai
phone: +1 (302) 566-5456
I enjoy learning new things, helping others learn new things, building products to help others learn new things, and building an organization that builds products to help others learn new things. After a research career spanning machine learning, human computation, and computational social science, I started Parsnip to build AI that helps people learn real-world skills, the way a good teacher would. I love solving interdisciplinary problems that sit at the intersection of technology, learning, and human behavior — the ones that need a bit of knowledge about everything.
Experience
Parsnip
2020—present
Co-founder and CEO
Parsnip lets users “level up” their real-life cooking with a competency-based
personalized learning system that can teach any skill to any learner, any time.
Our learning system generalizes to other domains such as K-12, higher education,
and workforce training, serving as an AI-powered “learning infrastructure” that
facilitates personalization for almost any learning product.
CTRL-labs (acquired by Facebook)
2017—2019
Research Scientist & Product Manager (New York, NY)
Designed a machine learning platform for organizing 1TB+ of time series datasets
& TensorFlow models, and automating model performance benchmarks on test data.
Prototyped 2D and 3D interaction applications for demonstrating user interface
control via machine learning from electromyography data.
Led a cross-team effort to build cloud infrastructure for training user-personalized machine learning models. Product manager for momentary event/signal detection from electromyography data, including R&D, data collection, accuracy metrics, user interface design, implementation, and product release.
Aarhus University
2017
Assistant Professor, tenure-track (Aarhus, Denmark)
Developed experiments and analyzed behavioral data on the ScienceAtHome
citizen science platform.
Microsoft Research
2015—2017
Postdoctoral Researcher, Computational Social Science (New York, NY)
Ran the longest experiment of iterated prisoner’s dilemma ever
conducted—lasting one month. Studied teamwork and collective problem
solving in groups of various size. Developed a web framework for
conducting real-time experiments of collective behavior using
participants from the Internet.
2013, 2014
Research Intern, Computational Social Science (New York, NY)
2012
Research Intern, Adaptive Systems and Interaction (Redmond, WA)
Developed improvements to crowdsourcing systems by building online predictive
models of user engagement and measuring the effects of
microeconomic and social incentives.
Education
Harvard University
Ph.D., Computer Science
2015
Dissertation: Experimental Studies of Human Behavior in Social Computing Systems
S.M., Computer Science
2012
University of Pennsylvania
B.S., Economics, summa cum laude; The Wharton School
B.S.E., Computer Science, summa cum laude (GPA: 3.89 / 4.0)
2009
Qualifications
Scientific Research: 12 peer-reviewed publications with over 1,200 citations (h-index: 12) in top computer science conferences (AAAI, EC, HCOMP) and interdisciplinary journals (Nature Communications, PLoS ONE) spanning artificial intelligence, human computation, and computational social science. 31 invited talks at universities, research labs, and conferences worldwide. Program committee member for over 8 international conferences in computer science and computational social science, including WWW, IJCAI, AAAI, AAMAS, EC, HCOMP, and ICWSM. For details see my CV.
Open-source: TurkServer, a real-time web platform for conducting online experiments of collective behavior. CrowdMapper, a collaborative crisis mapping web app for studying teamwork and collective intelligence (see talk).
Digital Fabrication: CAD/CAM CNC machining, microcontroller programming and circuit design, laser cutting, 3D printing, casting and composites. See my work from MIT’s How to Make (Almost) Anything.