Ziad Yassine, PhD

• Designed and executed a discrete choice experiment to quantify consumer adoption of drone delivery versus courier delivery (e.g.,Uber Eats,DoorDash) as a function of cost, time, and individual behavior.
• Developed a market segmentation model using discrete choice analysis to identify key consumer personas and predict drone delivery adoption based on demographics, behavioral factors, and operational characteristics.
• Orchestrated a controlled crossover experiment to estimate drone delivery impacts on customer satisfaction; conducted randomized trials and analyzed repeated measures to reveal a 30% increase in overall satisfaction.
• Simulated supply chain operations for medical product deliveries by developing geospatial models to optimize drone and truck delivery routes, enabling a hybrid system that improved energy efficiency by 50%.

Led a behavioral and environmental impact analysis of an electric vehicle carsharing system:

• Implemented a finite state machine to transform trip data into a vehicle state tracker, enabling the inference of vehicle charging activities and associated emissions, revealing a 43% net emission reduction from electrification.
• Modeled user retention using survival analysis, showing a 25% increase in retention due to a 20% pricing incentive.
• Developed a geospatial framework that models user access to grocery stores in terms of cost, distance, and time.

• Facilitated in-class discussions and office hours for over 100 students, graded assignments, and provided personalized feedback, achieving a 6.8/7.0 instructor rating and exceeding the department's average rating of 6.4.

• Evaluated innovative mobility on demand pilots in collaboration with the U.S. Department of Transportation.

• Developed an AI-driven personalized meme generator leveraging BERT's word embeddings and OpenAI's GPT model, and engineered a Python-based pipeline to process user requests in real-time.

• Developed a time-series regression model using Uber Movement data to forecast COVID-19 lockdown impacts on traffic speeds, leveraging spatial-temporal features to detect distributional shifts and improve model adaptability.

• Engineered text-based features like text length, punctuation usage, and capitalization patterns to train a logistic regression classifier for email spam detection, achieving over 88% accuracy.