New York, New York, United States
Account Abuse Team
• Designed and shipped a ‘Gibberish Name Detection’ feature for fraud detection, handling name-specific edge cases to complement existing email-based signals; contributing to a ~4% increase in ROC-AUC for top enterprise customers
• Drove end-to-end deployment of a DomainMX ML feature, from requirements and technical scoping through data ingestion, model integration, and production A/B validation
• Owned monthly ensemble tuning for 20-30 enterprise customers, evaluating optimal combinations of XGBoost, logistic regression, and global vs. per-customer models ahead of each production release
Content Integrity Team
• Led backend development for ‘Text Clusters’ and Content-Based Elasticsearch systems, supporting 150+ enterprise customers and millions of monthly users
• Improved stability and ROC-AUC of ML-based content scoring through monitoring score drift and conducting ensemble model experimentation ensure consistent customer impact across customers
• Authored technical design documents for service migrations and Console improvements, and served as a first-line incident responder for critical production services