At Cellula Technology, I built end-to-end hotel booking cancellation and Uber fare prediction pipelines, then integrated each model into a FastAPI REST API. The projects achieved 87% accuracy and a 0.91 minority-class F1-score for cancellation prediction, and an R² of 0.74 for fare prediction.
For my university graduation project, I built a real-time industrial quality control system that detects products and defects in video, reaching validation mAP50 scores of 99.4% and 94.2%. I’ve also developed prediction systems for taxi trip duration and road accident risk, and a deep learning model for estimating body measurements.

