My Projects
Welcome to my portfolio! Here, you can explore the various projects I've worked on. While I can provide descriptions for the major business projects I've worked on, I cannot provide exact specifics or visuals.
Enterprise Worker's Compensation Pricing Model
My Role: Co-Primary Data Scientist
For this project, I was one of the primary data scientists who developed the new enterprise-wide worker's compensation pricing model. This model brought together data from 4 different internal brands, different rating bureaus, and other external sources to create the first unified pricing model.
As one of the primary data scientists, I was one of those responsible for analyzing the incoming data, implementing data cleaning techniques, training model iterations, and evaluating the final results. This project was successfully deployed and was in use with only minor updates for over half a year when I left AF Group.
Worker's Compensation Claim Litigation Models
My Role: Senior Data Scientist
I was responsible for the development of 2 suites of models which evaluated the probability of worker's compensation claims evolving to different litigation outcomes.
The final model suite deployed to production included over 30 models which handled different claim scenarios. The models helped our claims staff better allocate resources and earlier identify problematic claims.
Medical Malpractice Insured Claims Predictor
My Role: Primary Developer
For this project we sought to utilize internal account information, along with some external data, to predict the probability that a given insured would have a malpractice claim within the next 3 and 5 years.
Using various software packages, I created the underlying data pipelines, model scoring pipelines, and trained the final models used to predict the claims probabilities. These models were combined with an Underwriting model to provide insights at binding for our Underwriters.
Identifying False Positive Planet Detections from the Kepler Space Observatory
My Role: Lead Researcher
One of the final projects I worked on during graduate school was performing follow-up observations of potential exo-solar planets. As the lead for this project, I wrote the proposal for time as part of the SDSS-III survey, identified and provided the relevant targets, and performed the data analysis.
As part of this project, we observed over 150 potential planets to identify false positives and gain a better understanding of these targets' population. As part of this project, I automated an existing IDL pipeline and increased the data preparation throughput by 10x.
To the right is the first identified false positive I made during this project. For a time, I was probably the only person who knew that this second star existed which was very cool to think about.
Reducing False Positive Planet Detections by Identifying Stellar Noise Contributions
My Role: Lead Researcher
My second published scientific paper was written from the results of my second year project. For this project, we used a novel new technique to identify noise from stars and combined it with data from a NASA solar observatory satellite.
Combining these, we were able to use the sun as a proxy and identify the wavelengths of light which produced the least amount of stellar noise. These observations could allow future instruments to be built that would reduce the chance of false positive planet detections.
Mapping Magnetic Fields in Galactic Molecular Clouds
My Role: Lead Researcher
As an undergraduate, I worked as part of the Galactic Plane Infrared Polarization Survey (GPIPS). My main research, and what later became my first published scientific paper, focused on combining the survey's data with data from 3 other surveys to create the first resolved map of magnetic field strengths in a molecular cloud.
These maps gave the first estimated strength of magnetic fields in molecular clouds. These measurements were important because up until these measurements, astronomers used rough estimates when simulating star formation. Now, we have an accurate method to measure these values which will help us better understand how stars are formed and what role the galactic magnetic field plays in that process.
