Welcome to My Personal Website

Senior Manager (Lead Model Developer) at Bank of Nova Scotia

June 2022 - Present

Spearheaded the development of customer analytics and insights strategy, and commercial rating models with End-to-End Modeling Leadership.

• Designed and implemented the bank's first-generation first-party fraud detection models for Caribbean credit card and unsecured loan portfolios, achieving 10x to 50x lift in fraud identification.
• Redesigned and migrated the bank's comprehensive Caribbean portfolio review dashboards from Tableau to Power BI.
• Led data-driven strategy design to enhance risk-reward predictions, customer segmentation, credit limits assignment, and risk-based pricing.
• Redeveloped multiple credit rating models using machine learning techniques such as Random Forest and XGBoost, optimizing model weights using rank-based correlations and genetic algorithms.

Award: Honored with the "Best of Best 2023" performance award for outstanding contributions.

Senior Data Scientist at Deloitte

July 2018 - June 2022

Worked on diverse projects in User Behavior Prediction, Credit Risk and Market Risk Models.

• Developed first-generation advanced attrition and points redemption models for credit cards, leveraging discrete event-history modeling and feature engineering.
• Developed Probability of Default (PD), Loss Given Default (LGD), and Exposure at Default (EAD) models for regulatory compliance (IFRS 9).
• Audited several retail portfolio credit risk rating models, assessing their validity and regulatory compliance.
• Validated pricing models for bond futures, interest rate futures, interest rate swaps and risk rating models for asset-backed securities.

Model Validation Specialist at Bank of Montreal

November 2016 - June 2018

• Validated multiple retail product stress testing models (including PD, LGD, EAD, PPNR, and fair lending).
• Led the validation of a pioneering Credit Card PD model built using an account-level hazard rate methodology.
• Contributed to the development of model monitoring frameworks; designed metrics (KPI) and established thresholds.

Biostatistician at Princess Margaret Hospital

May 2010 - August 2011

• Applied dimensionality reduction algorithms to perform feature selection on data with extremely low signal-to-noise ratios.
• Built diagnostic and predictive time-to-event survival analysis models on head and neck cancer patients.

Statistical Analyst (Co-op Intern) at University of Alberta

May 2009 - December 2009

• Applied cluster-effect-adjusted non-parametric tests and longitudinal data modeling to identify risk factors.
• Performed time series analysis and built multiple regression models for various medical research projects.