• Created application for bicyclists that provides safer routes, avoiding known dangerous intersections.

• Implemented random forest to model traffic patters and empirical Bayes model to estimate the probability of a collision using public traffic studies, accident data, and OpenStreetMaps network.

• Provided routes that were 68% safer than the fastest route on average.

Built convolutional neural networks to detect smuggling boats. Implemented YOLO and Faster-RCNN network architectures for real time detection in Tensorflow, achieving 80% accuracy.