At Carey International, I developed Holt–Winters models for city-level trip volume that achieved 12% MAPE on rolling backtests. I evaluated forecast-driven fleet recommendations against manual planning to support capacity planning and reduce under-capacity risk.
I also developed RFM-based customer segments and Ideal Customer Profiles across 7 segments and 470K+ annual trips, increasing qualified accounts identified by approximately 26%. My analysis of trends across 2,000+ corporate accounts surfaced growth, decline, and revenue opportunities.
At Bahwan CyberTek, I developed and deployed a Retrieval-Augmented Generation chatbot for real-time data queries. I also built demand forecasting models that reduced MAE by 15% over the baseline and deployed them as Flask REST APIs on AWS.
At Calix, I developed an XGBoost classifier that improved F1-score by 22% over baseline and built Looker dashboards for subscriber segmentation. As a Data Science Intern at RAPP, I worked on ROAS prediction and marketing mix modeling, including data pipelines that reduced processing time from 3 minutes to 45 seconds.
