Projects (Work and Personal)

The Washington Post

Voterflow Inference

I developed a stratified regression model to estimate the share of a candidate’s votes in a prior election that is going to a new candidate in a current election.
pandas cvxpy PyMC Plotly matplotlib

Polling Averages

I co-developed our Bayesian Presidential polling averages, which generated over 9 million pageviews and was the most-read Politics story in 2024.
pandas NumPyro scikit-learn AWS Batch MySQL SQLAlchemy

Live Election Model

I contributed to our model that predicts who will likely win elections as votes are tabulated live. You can find the code for our Live Model on Github.
pandas cvxpy NumPy SciPy

Educational Testing Service (ETS)

e-rater

I was a key member of the team responsible for re-architecting and developing ETS’s automated essay scoring platform. We transformed the existing code-base into a modular NLP feature- and feedback-generating engine using Apache Storm.

Learn more

TextEvaluator

I transformed TextEvaluator from a Perl script into an object-oriented, extensible Python program. TextEvaluator combines NLP features via regression to deliver the most accurate measure of reading complexity and genre-classification available. Its readability measurements have been used by McGraw-Hill Education.

Learn more

Side Projects

2025 New Jersey Governor's Race

I’m working on Bayesian polling averages to infer who will likely become the next Governor of New Jersey.
pandas PyMC matplotlib seaborn