☰

Data Science

My quantitative background is in computational neuroscience. During my days as a neuroscientist, I spent most of my time analyzing time series data, so I'll always have a soft spot in my heart for things like signal decomposition and forecasting models.

Nowadays, I like to apply my skillset more widely. Here are some topics I'm interested in:

I'm obsessed with learning more about knowledge graphs.
Maps. Maps of humans, of trees, of butterfly migrations, of limestone, of clown attacks, etc.
I'm passionate about applying my skillset to the healthcare and e-learning industries.
I have a background in human factors, so I care deeply about usability and explainability in AI.
I love a good UI.

You can find my resumé and a list of projects below.

download resumé

A collection of recent projects

SugarTime
The SugarTime package is a set of tools to model the effects of carbohydrates and insulin on blood glucose dynamics. The goal of these tools is to learn how a diabetic's blood glucose responds to food and to insulin. After it has learned this, the patient can provide the timing and amount of carbs they plan on eating, and the model will offer a best guess as to which insulin dosage will keep their blood glucose within an optimal range.
SearchMo
This started off as an exploration of my Mormon roots. It is currently an app that provides N-Gram frequency search functionality for a corpus of sermons and lessons from LDS church leadership. It covers from the mid 1800s to today. I have plans to expand into other religions/traditions.
DuraScrape
DuraScrape is a web crawler that scrapes open-access journal articles from academic journal websites and saves them in a SQL database.
NERF:
Nonlinear Estimation of Receptive Fields
This is currently a repository of MATLAB and python code for preprocessing and analysis of behavioral and neural timeseries data. Beyond I/O, cleaning, and basic plotting functions, it specifically provides tools for inferring the spatiotemporal preferences of visually responsive neurons.
vis3DConnect
Vis3DConnect is a python toolset for 3D visualization of brain connectivity data. It allows the user to select neural tracing experiments from the Mouse Brain Connectivity Atlas at the Allen Institute, and to visualize them in a 3D glass brain with an interface that allows rotation, zooming, etc.