Kingston, New York, United States
• Trained and deployed a YOLOv11 PyTorch model with OpenCV and GStreamer to track vehicles in real time, reducing reliance on $15k laser-based detection hardware.
• Integrated PyTorch YOLOv11 with OpenCV and GStreamer in a multithreaded Python application to track vehicles in real time and convert pixel detections into road coordinates, eliminating hardware-based size estimation.
• Developed Golang-based UDP trigger system with MySQL stored procedures, reducing storage costs by capturing images only when vehicles are detected by the machine learning algorithm.
• Optimized ML algorithms for Intel CPUs using OpenVino, eliminating GPU dependency and reducing hardware costs.
• Implemented Zabbix monitoring and email alerts for servers and IP cameras, increasing system reliability.
• Designed REST APIs to access vehicle data, enabling integration with front-end applications.
• Built multi-lane traffic simulator streaming data via 16 WebSockets, accelerating front-end testing in the absense of real
road data.
• Converted scripts into Linux system services with auto-restart and log rotation for the production environment
• Authored technical documentation and Visio diagrams for future developers