I'm a graduate of UChicago's Computational Analysis and Public Policy program. I have a passion for data science and experience in the consulting, financial services, and public sectors. My most recent work is as an AI strategy consultant and data science contractor.
I previously worked as a Data Science Research Assistant at UChicago's Crime Lab, where I supported the evaluation of randomized controlled trials by using natural language processing on video data. Before that, I led the Data Team at Financial Health Network as a Senior Manager.
You can see a few of my projects at the links below:
-
Civic Lens website. This open source site makes federal public commenting opportunities more accessible for the American public. I am lead backend engineer of this ongoing project, working with Django, PostgreSQL, and Python to manage the data pipeline for the site. Our code lives here.
-
Predicting Observed Poverty Levels in Costa Rica. Alongside three teammates, I tested different machine learning models, as well as techniques like ensemble learning, oversampling, and cross-validation to classify household-level poverty using observed characteristics. Our final model approached the top performing Kaggle models with a final Macro F1 score of .42. We summarized our approach and results in a summary report.
-
Comparing a Hardship Index with Evictions in Chicago. Using publicly available data, three teammates and I constructed an dashboard using Python and plotly to compare a constructed hardship index with instances of eviction. The resulting interactive data visualization allows the user to examine the index, its data inputs, and eviction data at the zip code-level in Chicago. We summarized our work in an explanatory report.
-
A/B testing of messaging in a fintech app. I designed and implemented A/B testing of behavioral design-informed messaging in a fintech app, resulting in 31% higher enrollment in automated savings program and 42% greater savings balances for users.
Other (private) projects include: building an app for a national nonprofit using an LLM to automate previously manual text extraction from hundreds of documents, constructing a data processing pipeline and testing different machine learning techniques to improve the accuracy of predicted policing outcomes by 40%, and improving a customer classification model for a B2B fintech company.
Feel free to connect with me on LinkedIn!