At Theory+Practice, I build data products for Continuum AI and consulting clients, turning forecasting, optimization, and risk models into business actions. I productized a demand-forecasting and genetic-algorithm price/promotion optimization feature that achieved up to 96.5% forecast accuracy and delivered an average 8.1% promo lift across 23 markets.
I've built an XGBoost disruption-risk model for Johnson & Johnson MedTech, identifying approximately $3M in revenue at risk in a quarter, and a LightGBM forecasting model for L'Oréal that surfaced approximately $30M in recoverable opportunity. For Perkopolis, I led cross-functional targeting work that increased open rates by 30% and click-through rates by 60%.
My foundation combines applied data science with optimization research: at UBC, I developed a bi-level edge-computing service placement and pricing framework published in IEEE IoT Journal. Earlier, at Capgemini, I built SAP data flows, reporting objects, and validation processes for reliable analytics.
