Work

Module Federation

I led the migration from a legacy custom forms engine to a Webpack Module Federation architecture across multiple Angular repositories. The goal was to replace a fragile, poorly documented system with independently built Angular apps that could be consumed by a shared host platform.

This work involved building migration templates, supporting hybrid legacy/federated apps, coordinating with backend and deployment teams, debugging AWS artifact issues, and resolving a production-impacting Webpack chunk race condition. The result was a faster local development workflow, easier onboarding for new developers, and a more maintainable path away from the legacy forms engine. Read more.

Billboard Data Analysis

I created a data science project analyzing sociolinguistic variation in Billboard #1 songs from 1940 to 1970. The project involved building a custom research dataset from historical chart data, cleaned lyrics, biographical metadata, and manually coded audio observations.

On the technical side, I wrote Python scripts using NLTK to identify potential instances of a target lexical set in song lyrics and generate annotation templates for manual coding. I later turned this workflow into a small web application with a React front end and a Python API hosted on PythonAnywhere, making it easier to generate templates, count lexical-set occurrences, and adapt the tool to additional variables.

After collecting the data, I visualized changes in the sociolinguistic variable over time and built a scikit-learn regression pipeline to explore whether year, race, and sex could predict variation in pronunciation. The final multiple linear regression model outperformed a median baseline and explained about 20% of the variation in the held-out test data. Read more.