Updated periodically

Machine learning & NLP

  • #democracy: trying to predict public opinion polls in the 2020 Democratic primaries using Twitter, NLP, and Bayesian modelling. Tools used: AWS, TWINT (Python package), Keras.
  • Using topic-controlled document vectors to measure the ideological slant of news media: A summer research project at the University of Toronto where I built a machine learning model to detect the ideological slant of news articles. Tools used: crowdsourcing, word vectors, neural nets. Libraries: spaCy, Keras.

Data Visualization

  • Normalizing flows: A visualization of normalizing flows, a technique for transforming simple distributions into more complex ones for use in variational inference. From Rezende & Mohamed, 2015. Made with D3.js
  • Marvel Cinematic Universe: Exploring the success and dominance of the Marvel Cinematic Universe film franchise. Made with D3.js
  • Mapping COVID-19 Contact Tracing Applications Around the World: A visualization and database of contact tracing applications around the world for COVID-19. Hoping to keep this updated frequently. Made with D3.js
  • Police Reform Tracker: An interactive map and table showing polie reforms passed by state legislatures across the United States. I was part of a team of two devs and many more people keeping track of laws and developments in the fight for police accountability. Made with react.


  • COVID-19 Student Response Team: The website of the Harvard Medical Students COVID-19 Response Team – a group of HMS students producing and disseminating educational resources, as well as coordinating activism initiatives surrounding COVID-19. Made with Hugo