Experience

Buoy Health

• United States
• Wellness and Fitness Services
• 1 - 100 Employee

• Senior Machine Learning Research Scientist

  • Jan 2021 - Present

  Buoy builds a free digital health tool that helps people – from the moment they get sick – start their health care on the right foot. Started by a team of doctors and computer scientists working at the Harvard Innovation Laboratory in Boston MA, Buoy was developed in direct response to the downward spiral we’ve all faced when we attempt to self-diagnose our symptoms online. Buoy leverages artificial intelligence – powered by advanced machine learning and proprietary granular data - to resemble an exchange you would have with your favorite doctor – to provide consumers with a real-time, accurate analysis of their symptoms and help them easily and quickly embark on the right path to getting better. Show less

• Machine Learning Research Scientist

  • Jan 2018 - Jan 2021



Insight Data Science

• United States
• Higher Education
• 1 - 100 Employee

• Artificial Intelligence Fellow

  • Sep 2017 - Dec 2017

  Built a cat photo analyzer as a marketing tool for a collaborated company, Snappr. - Trained and tweaked a Inception-Resnet model that can identify cat breed among 42 classes, as well as the alertness, cuteness and emotion of a cat. - Achieved 90% top-1 accuracy and 98% top-3 accuracy in classifying cat breed. - Utilized Amazon Rekognition API to prevent people from tricking the model with non-cat pictures. - Delivered an engaging, fun web app (Snapcat) for people to play with. Built a cat photo analyzer as a marketing tool for a collaborated company, Snappr. - Trained and tweaked a Inception-Resnet model that can identify cat breed among 42 classes, as well as the alertness, cuteness and emotion of a cat. - Achieved 90% top-1 accuracy and 98% top-3 accuracy in classifying cat breed. - Utilized Amazon Rekognition API to prevent people from tricking the model with non-cat pictures. - Delivered an engaging, fun web app (Snapcat) for people to play with.



Massachusetts Institute of Technology

• United States
• Higher Education
• 700 & Above Employee

• Research Assistant

  • Sep 2012 - Aug 2017

  Modeled the thermodynamic behavior of block-copolymer. - Developed a convolutional neural net model that recognizes three different copolymer patterns with the accuracy over 99%. - Performed image analysis and measured the pattern quality of each experiment. - Built a regression model that predicts the pattern quality with given time-series experimental parameters, which helps us find the optimal experiment setup and reduce the pattern defect density by ~20x. Developed scalable lithography method at 10 nm node for future magnetic devices such as bit-patterned media and magnetic random access memory, and exploring the nanoscale magnetism in a statistical manner. Show less



National Taiwan University

• Taiwan
• Higher Education
• 700 & Above Employee

• Graduate Researcher

  • Sep 2009 - Jun 2011

  Originated and illustrated the idea of all-carbon based solar cell that generates energy in a sustainable way. Originated and illustrated the idea of all-carbon based solar cell that generates energy in a sustainable way.