Amir

Mehdizadeh

Architected and own enterprise-scale forecasting systems serving 8,000+ hotels across 120+ countries, powering revenue optimization, capacity planning, and financial forecasting with daily automated predictions. Built probabilistic and hierarchical forecasting frameworks with Bayesian methods and optimal reconciliation algorithms, enabling Finance teams to quantify revenue risk across property, market, and regional levels. Addressed cold-start challenges for new hotels via similarity-based synthetic data generation, and led model selection across 10+ algorithms using an automated backtesting framework.

During my PhD, I led three applied research projects spanning safety and energy analytics. I built an ML framework (XGBoost, neural networks, SVM) to predict dangerous driving events 30 minutes ahead using 1.2TB of fleet IoT data, achieving 76% AUC, and optimized vehicle routing with crash risk constraints. In manufacturing, I developed a job rotation optimization model using fatigue-failure theory, reducing predicted injury rates by 25%. I also built a BI platform for Southern Power Company to identify profitable energy trading strategies in the ERCOT market using time series forecasting and Power BI dashboards.