Blake Atkinson
Data Scientist
Fort Thomas, Kentucky, United States
Summary
👤
Senior🎓
Top SchoolBlake Atkinson is a data scientist with eight years of experience, blending probabilistic modeling and graph-based methods in sports analytics and beyond. He is currently a Data Scientist at Angstrom, where he applies advanced ML to practical data-driven decision making. Previously at G Street Analytics, he automated data collection from diverse sources, engineered features, and built time-series, tabular, and graph models with automated validation checks. As a freelance developer, he built end-to-end horse racing prediction pipelines, earned a top-3% finish in the NFL Big Data Bowl, and delivered multiple web solutions. He holds a BA in Physics from Transylvania University and an MS in Applied Data Science from the University of Michigan, equipping him with strong quantitative foundations. Notably, he contributed the Poisson distribution and related components to NGBoost—the probabilistic boosting library from Stanford ML Group—demonstrating practical open-source impact with SHAP-based feature analyses.
8 years of coding experience
6 years of employment as a software developer
Master of Science - MS, Applied Data Science, Master of Science - MS, Applied Data Science at University of Michigan - School of InformationBachelor of Arts - BA, Physics, Bachelor of Arts - BA, Physics at Transylvania University
Github Skills (16)
scikit-learn10
scikit10
data-modeling10
machine-learning10
probabilistic-programming10
statistical-modeling10
statistical-modelling10
python10
numpy10
probabilistic-reasoning10
probabilistic-models10
gradient-boosting9
pandas9
matplotlib8
shap8
Programming languages (3)
Github contributions (5)
stanfordmlgroup/ngboost
Jul 2020 - Aug 2020
Natural Gradient Boosting for Probabilistic Prediction
Role in this project:
Data Scientist / ML Engineer
Data Scientist / ML EngineerContributions:9 commits, 1 PR, 3 comments in 1 month
Contributions summary:Blake primarily contributed to implementing the Poisson distribution within the NGBoost framework. Their work involved adding the `Poisson` class, implementing `PoissonLogScore`, and defining the `fit` and `sample` methods. They also created an example demonstrating the use of the Poisson distribution within NGBoost, including feature importance analysis and SHAP plots. The user's contributions demonstrate a strong understanding of probabilistic modeling and machine learning concepts.
pythonpredictionnaive-bayesnatural-gradientsmachine-learning
btatkinson/golf_scraper
Apr 2019 - Jun 2019
Contributions:15 commits, 14 pushes, 1 branch in 2 months
Find and Hire Top DevelopersWe’ve analyzed the programming source code of over 60 million software developers on GitHub and scored them by 50,000 skills. Sign-up on Prog,AI to search for software developers.