• Spearheaded development of ML models to replace rule-based systems detecting suspicious activity as part of Anti Money Laundering (AML) team
• Worked on predictive models for Small Business and Consumer Card portfolio spanning over a 100 million accounts across 80 million customers
• Employed a data-driven strategy to engineer 300+ features that identify risky transactions, improving models’ ability to detect subtle and evolving risks
• Demonstrated proficiency in Python, SQL and latest ML libraries and used supervised and semi-supervised classification algorithms to build the model
• Integrated SHAP and LIME techniques for model interpretability and explainability of model features ensuring compliance with FinCEN guidelines
• Results suggest 40% alert reduction with an estimated savings of $50 million at 70% recall, along with 30% net new alerts covering unknown/emerging risk
We will review the reports from both freelancer and employer to give the best decision. It will take 3-5 business days for reviewing after receiving two reports.