I'm comfortable working with Python (pandas, polars, numpy, matplotlib, PySpark, scikit-learn) and R (tidyverse, brms). I have experience writing SQL queries in a variety of flavors (BigQuery, AWS Athena, SnowSQL, DuckDB). I also have experience writing shell scripts and Lua for Neovim plugins.
I was first author on a paper analyzing the early days of the COVID-19 vaccine rollout in the US. This involved a clustering analysis to identify counties of low vaccine uptake and building county-level logistic growth models (using a Bayesian approach for fitting) in R to describe patterns in vaccination coverage over time. I also designed, built, and deployed a Plotly/Dash dashboard to track county-level vaccination progress. It reached a peak of nearly 20k users in July of 2021.
I’m currently working on a project focused on modeling the seasonality of flu cases in the US. I’m using AWS Athena and SageMaker to build datasets of confirmed flu counts at various spatiotemporal scales based on medical insurance claims data.
With the help of the Massive Data Institute at Georgetown, I coordinated the lab’s adoption of Google Cloud Platform (GCP) for working with our medical insurance claims data. I learned how to use pywinauto to automate the decryption of the data files from their proprietary form and used Dataproc/PySpark to convert the data from CSV to Parquet. Later, I researched and pushed for a shift towards BigQuery after realizing it would better suit our use-case in the long-term. We’re now able to move more quickly and iterate on queries more efficiently.
I wanted to find a way to make on-the-fly settings overrides in Wezterm (my terminal) for things like the background, tab bar, or font size from within Neovim. Otherwise, I’d have to manually edit the config file. This would be useful when sharing your screen in a video call or cutting down on visual distractions during focused work. I built a plugin in Lua that accomplishes this. It involved finding the right terminal escape codes and figuring out how to use Neovim’s powerful API to encode and send Lua tables as JSON, then use that JSON data inside of Wezterm to apply the actual changes.
