Bryan S.

MLE focussing on the Venmo product. Projects include personalization models, churn prediction, and friend request recommendations

DS lead for the peer to peer ease of payments team Lead a small team (3 people) focusing on experimentation, modeling, analytics, data pipelines, and reporting Responsible for development of experimentation tools for all peer to peer DS teams Owned instrumentation and data pipeline for shopping analytics team

• Built a data pipeline to analyze LinkedIn connection requests to identify fraudulent connections from recruiters. Partnered with product managers to define what it means for a connection to be fraudulent • Conducted a causal study to identify the impact of receiving at least one InMail and engagement on LinkedIn. Determined a significant relationship exists, contributing to a department-wide project to increase effectiveness of InMails and user engagement • Built a data pipeline using Scala to match free text to standardized entities. Proved the need for standardization of keywords recruiters use to search for candidates • Designed A/B tests on the impact of new features. Conducted power analyses and presented results to product team

• Used NLP on consumer reviews and complaints to identify key themes and pain points for a payments company. Used to inform a company plan to address pain points and highlight positives in marketing messaging • Built data pipelines and machine learning models to guide call center employees on the next best action to increase customer retention and cross-sell. Implemented in production and provides regular recommendations, resulting in annual savings of $10MM. Used R, Python, Driverless AI, and SparkBeyond • Built ETL pipeline in Python using AWS to perform quality control and standardize disjointed data from 500 companies for the 2021 Women in the Workplace report. Decreased time to standardize data from weeks to hours. Removed the need for a data engineer to assemble future reports, even as data formats change • Built a machine learning model for a medical client to ensure patients receive the care they need. Combined insurance claims data from multiple sources. Reconciled differently formatted IDs, missing data, and procedure naming conventions. Conducted research with SMEs and codified rules previously governed by doctor intuition

Client facing data scientist for private sector operations within the Advanced Analytics practice.

Client facing data scientist for private sector operations within the Advanced Analytics practice.

Data Scientist - Internal Revenue Service fraud detection project • Decreased processing time for dataset pipeline by 50% and processing time for filters by 30% by optimizing SQL queries and addressing technical debt • Built a predictive model using SAS to identify first-person tax fraud, resulting in $40MM in savings per year • Integrated predictive models into a production system using SAS Embedded Process Technology

Wrote analysis files in STATA and developed proprietary excel spreadsheets for litigation reports involving multinational banks Wrote STATA programs to clean and prepare datasets for mortgage discrimination analysis Updated mortgage redlining analysis to comply with new census track data Applied interest rate discrimination analysis to several litigation cases Linked ArcMap census data to STATA crosswalk programs to allow for easy mortgage redlining analysis Completed summer internship and was invited to stay for fall semester Represented Charles River Associates at the GWU Careers in Economics event