Zachary Jones
Senior Applied Scientist at Amazon
San Diego, California, United States
Summary
🤩
Rockstar🎓
Top SchoolZachary Jones is a Senior Applied Scientist in San Diego with 13 years of experience building production ML systems that detect fraud and adversarial behavior at scale. He has led end-to-end model development and deployment at Amazon and Meta, turning experimental insights into real-time decisioning and enforcement systems with tens to hundreds of millions in business impact. His background blends rigorous academic research (a PhD and postdoctoral work producing widely used interpretable ML tools and peer-reviewed publications) with hands-on engineering—maintaining complex pipelines, online drift monitoring, and model calibration in sensitive subgroups. An active contributor to the mlr machine-learning ecosystem, he has improved plotting, missing-data handling, and feature-importance methods for practical model validation. Notably, he pairs political-science expertise in causal and network bias problems with production ML tooling, making him particularly strong at measuring and mitigating adversarial and latent-confounder risks.
13 years of coding experience
5 years of employment as a software developer
Doctor of Philosophy - PhD Political Science (Statistics and Computational Methods), Doctor of Philosophy - PhD Political Science (Statistics and Computational Methods) at Penn State University
Github Skills (8)
ggplot10
mlr10
data-science10
r10
classification9
machine-learning9
regression9
survival-analysis8
Programming languages (6)
Github contributions (5)
mlr-org/mlr
Mar 2015 - Feb 2018
Machine Learning in R
Role in this project:
Data Scientist & ML Engineer
Data Scientist & ML EngineerContributions:112 commits, 125 PRs, 95 pushes in 2 years 11 months
Contributions summary:Zachary focused on enhancing the machine learning capabilities of the mlr package, addressing missing data handling in conditional inference trees (cforest), and by adding support for plotting ROC curves using ggplot2 and functions from the plotROC package. The user also refactored the code to separate the plotting and generation of ROCR curve data and added a new feature importance filter based on the party package's cforest model. Furthermore, the user introduced several new plotting functionalities for threshold vs. performance curves and learning curves by leveraging ggvis.
machine-learningrdata-sciencetuningcran
cdcrabtree/causalinference
Nov 2013 - Mar 2015
Contributions:25 commits in 1 year 4 months
sciencecausalinferencepoliticalmachine-learning
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.