-Refactored code for training purchase prediction models from Scala to Python resulting in a 10x size smallercodebase, 2x faster processing time, and a more maintainable solution.-Designed and ran hyperparameter tuning experiment that increased model performance from an AUCROCof 0.8 to 0.9 resulting in a more personalized search experience on Kohls.com.-Utilized pyspark and GCP tools like Dataproc and Vertex AI to build models and monitor production.

-Created PureSkill.gg, an AI Counter-Strike coach, telling players what they’re doing wrong and how to fix it.-Acquired more than 15,000 users and processed 1,000,000+ matches with an error/failure rate of under 1%.-Built pipeline to transform a game replay file to millions of usable points of data for each CS:GO match.-Designed 30+ player assessments that use machine learning, statistics, and other techniques to find mis‑takes and prioritize them based on importance. This effectively mimics a human coach for users.-Trained linear regression model on AWS Sagemaker to predict the probability of a team winning a roundwith an accuracy of 85%. This robustly finds low‑impact actions for players to work on.-Trained model using the clustering algorithm HDBSCAN to find commonly thrown smoke grenades, andcorrelated player rank with cluster to find incorrectly thrown smoke grenades.-Constructed browser 2D match viewer to seamlessly show players their own mishaps for easier correction.

Data Science consulting- Classified 100k medical documents every day using TF-IDF and cosine similarity - Developed recommender system for investors to find companies that match prior investments

Assortment Optimization leveraging agile methodology - Transformed large and complex datasets with SQL into usable formats for machine learning - Created several tools to make working with Lowe’s data more manageable. These are currently used by most of the data science team- Translated business requirements to data science objectives so that the product built will be intuitive, accurate, and satisfactory to the end users - Assembled and analyzed large datasets from third parties to determine their usefulness- Developed mathematical model behind the core algorithm

The precursor to PureSkill.gg. Independent project to analyze video game replays with machine learning, statistics, and AI in order to understand what data exists and can be extracted from CS:GO demos.- Collected several TB of video game replays for initial data analysis - Developed localized solution to extract data from game replays - Calculated how much impact a player’s actions had in each match with logarithmic regression to determine their contribution to the overall result (win or lose)- Tested various clustering methods to understand how players throw grenades and determine if a player is throwing a grenade correctly - Developed multiple algorithms to assess player choices when purchasing weapons and we will use user feedback to determine which is most effective

Discovered how exploding stars blow material out of galaxies, cause emission lines due to shocks, and impair the formation of stars - Collected data for 15+ nights on a 300+ million dollar, world class telescope as part of the largest survey ever conducted on that telescope- Created open source data processing pipeline to interactively extract useful, scientific data and created open source Markov Chain Monte Carlo fitting software to analyze galaxy light emissions- Invented, tested, and verified method to split galaxy emission signals into a galaxy component and an outflowing component to study the outflows independently of the host galaxy- Authored scientific paper describing results that is published in the Astrophysics Journal- Presented results to the general public through outreach events

- Designed, prototyped, tested, built, and flew a small (500 g) payload on a weather balloon to study the atmosphere up to an altitude of 100,000 feet - Selected and tested sensors to study the temperature, pressure, and humidity of the atmosphere as a function of elevation- Wrote and tested software to control payload during flight and to analyze the data after landing- Website with design documents, presentations, and data is listed under "Projects"

- Studied dust around red supergiant stars using the 3D radiative transfer code, Moccasin- Wrote extensive programs to analyze and visualize data outputs- Presented the work at the American Astronomical Society meeting in 2010- Won senior research award for LSU Physics and Astronomy Department- Awarded two LaSPACE Undergraduate Research Assistantships

- Made mixtures of olivine, pyroxene, and carbon to simulate the surface of asteroids - Took spectra of asteroids with the NASA Infrared Telescope Facility - Classified asteroids based on their spectra- Presented findings at the 42nd Division for Planetary Sciences meeting.

- Studied a binary star occultation of Saturn's rings taken by the Cassini spacecraft- Developed a method and accompanying software to disentangle both stellar signals- Demonstrated the first ever separation of signals from a binary star in an occultation of Saturn's rings- Presented results at the 2010 Division of Dynamical Astronomy meeting