At Sienz, I designed machine vision systems for multi-camera defect detection across industrial sorting machines. I scaled a fruit-sorting platform from 3 cameras to 9, eliminating the top-view blind spot while maintaining real-time throughput.
I also parallelized inference across YOLO models and developed an edge-deployed classification system that let field technicians create datasets, train and evaluate models, and deploy them on site. Previously, at MercadoLibre, I worked on fraud detection and internal model evaluation tools; at Plantium, I led development of a plant-classification and GPS-localization vision system.
