• Analyzed donor and alumni data in Salesforce CRM to produce insights that optimized engagement and event planning

• Designed and tracked data-driven outreach campaigns and created interactive dashboards to monitor KPIs, improving alumni and donor participation and enabling informed fundraising decisions.

• Fine-tuned RoBERTa on 349k+ forensic interview questions, classifying four legal question types and generating data-driven insights that improved research evaluation frameworks

• Engineered scalable Python/PyTorch pipelines to preprocess, tokenize, and batch-process 300k+ records, accelerating experimentation and ensuring robust model comparisons

• Diagnosed and corrected class imbalance via targeted oversampling and curriculum-style training, improving minority-class recall and reliability of downstream behavioral analyses

• Evaluated zero-shot and few-shot prompting strategies as alternative levers to reduce dependence on hand-labeled data, quantifying tradeoffs and informing next-phase model investment decisions

• Architected a centralized Data Catalog (Java/Spring Boot) to normalize 30M+ records monthly from 30+ banks, CRM and legacy ERPs, building ETL pipelines that established a unified source of truth

• Engineered a Server-Driven UI architecture (React, JSON Schema) to render dynamic dashboard and forms at runtime, abstracting tenant logic to reduce code duplication by 40% and eliminate custom client builds

• Scaled anomaly detection microservices to achieve 97% accuracy, driving 5 enterprise renewals and 20 new client

• Spearheaded a team to build a Selenium/Jenkins CI/CD automation framework, delivering 750+ tests 3 months early to reduce manual QA by 60% and block 35% of pre-merge bugs