Andrew Soncrant Email & Phone Number

Seattle, Wa, Us

Seattle, Washington, United States

Washington, District Of Columbia, United States

Washington, District Of Columbia, United States

• Lead model developer for transaction-level credit risk model.• Built data pipelines in spark for modeling datasets consisting of billions of rows of credit card transaction data.• Performed recursive elimination and hyperparameter tuning on hundreds of features and billions of rows.• Worked with software engineers to deploy final model into production environment where it must consistently score credit-card swipes in 30 ms or less.

Washington, District Of Columbia, United States

Machine Learning Operations Team (MLOps) November 2019 – July 2020• Wrote firm-wide whitepaper on the importance of well-formed machine learning training and inferencepipelines emphasizing the use of automated build processes (CI/CD), modular code, and containerization.Dept. of Defense (DoD) June 2019 – November 2019• Developed and deployed model drift detection algorithms in Python to alert DoD analysts to anomalies.• Wrote Python package for evaluating and comparing… Show more Machine Learning Operations Team (MLOps) November 2019 – July 2020• Wrote firm-wide whitepaper on the importance of well-formed machine learning training and inferencepipelines emphasizing the use of automated build processes (CI/CD), modular code, and containerization.Dept. of Defense (DoD) June 2019 – November 2019• Developed and deployed model drift detection algorithms in Python to alert DoD analysts to anomalies.• Wrote Python package for evaluating and comparing the performance of various object-detection models.• Developed components of software designed to autonomously tag incoming data streams and assess the performance of object detection algorithms in real-time.Veterans Experience Office, Dept. of Veterans Affairs October 2018 – May 2019• Facilitated the aggregation, storage, and analysis of over two million free-text survey responses from Veterans.• Automated the generation of numerous reports that are presented directly to the VA Secretary.• Developed tree-based XGBoost model and RShiny dashboard for demonstrating fraud-detection capabilities. Show less

Ann Arbor, Michigan, United States

• Taught lab sections for introductory statistics/data analysis course.• Managed instruction and grades for 100+ students at a time.

Dearborn, Michigan, United States

• Built ML pipeline in Python combining NLP with logistic regression to predict car loan default.• Analyzed tens of millions of call logs spanning several decades.• Obtained AUC of .85 predicting default using only text from calls to delinquent customers.

Master Of Science - Ms, Applied Statistics

Bachelor Of Arts - Ba, Applied Mathematics