top of page

DATA SCIENCE

I've been a data scientist in academia and industry, making statistical machine learning models for basic scientific research, healthcare, and e-commerce. These are a selected few works from industry -- please see adjacent work in my UX Research and Neuroscience pages, or reach out to me directly for more information.

Designing a new data curation method for business analysts

Year: 2021

Role: Data Scientist

Company Type: E-Commerce

​

Internal business analysts were in need of a simplified way to categorize their suppliers' products. This was historically challenging because each supplier had their own ways of describing, sorting, and labeling – a headache for an analyst who wants to find all of a single kind of product. Here I used a combination of unsupervised and active machine learning to build a tool that allowed analysts to quickly classify and label products regardless of the different labels across suppliers. The result was a more efficient and flexible method for labeling and organizing product data.

​

The images below have been modified for confidentiality.

DataCuration2.png

Abstracted illustrations of the old and new data labeling process

Improving model understanding for clients with Explainable AI

Year: 2020

Role: Data Scientist

Company Type: E-Commerce

​

Explaining why machine learning models give certain outputs can be difficult to understand, especially for folks without data science training. I was asked to help a client understand the outputs of one of our customer behavior models. To do this, I wrote a software package that leveraged standard explainability methods that we could share with clients, to improve their understanding, while at the same time offering protection of proprietary business information. Ultimately, this helped our client better understand how we used their data, and they in turn helped us improve our model.​

​

The images below have been modified for confidentiality.

General examples of explainable AI outputs

Drug-specific monographs for rapid-cycle analytics

Year: 2019

Role: Data Scientist

Company Type: Healthcare SaaS

​

This was my first major industry data science project after finishing my academic postdoc research. As part of the Sentry Data Systems and Agilum data science team, we set out to determine the feasibility of offering a service to help pharmacy and therapeutic committees determine the patient health outcomes through a scalable "medication use evaluation". I was tasked with preparing and analyzing electronic health records from thousands of patients across dozens of healthcare facilities, and building a prototype dashboard of visuals to help clinicians understand the effects of opioids and acetaminophen on pain and recovery in surgical patients. I worked with engineers, clinicians, and healthcare executives through an iterative data collection, statistical analysis, and design of visuals to ensure the results were robust and useful. This resulted in a prototype scalable medical use evaluation, the publication of which is below.

bottom of page