This report presents a real-world implementation of a recently developed traffic signal optimization system that uses a small percentage of vehicle trajectory data-also known as probe or telemetry data-as the only input. The system utilizes a new model for designing signal timing plans that leverages the temporal-spatial information of vehicle tra-jectory data as opposed to traditional software such as Synchro which relies on turning movement counts. This model allows us to accurately reconstruct average recurrent traffic states by aggregating sufficient historical data, even at low penetration rates. In the real-world deployment conducted in coordination with the Michigan Department of Transportation (MDOT) and the Road Commission for Oakland County (RCOC), researchers used vehicle trajec-tory data from an estimated 7% of vehicles in the traffic network to update cycle lengths, splits, offsets, and timing schedules at seven intersections along a 2.5-mile stretch of a coordinated-actuated arterial in Pontiac, Michigan. This corridor recently went through a traditional signal optimization based on vehicle count data, providing a unique opportunity to directly compare the performance of the new vehicle trajectory-based system against conventional methods. The vehicle trajectory-based system outperformed the traditional approach, reducing the overall control delay by 17.4% and the number of stops by 20.4%, compared to reductions of 13.7% and 14.9%, respectively, achieved by traditional optimization, and a cost analysis demonstrated savings of up to 35%. By utilizing increasingly available vehicle trajectory data as the only input and not requiring any additional infrastructure, we believe our sys-tem will provide a more scalable and economical solution to traffic signal optimization that could be applied world-wide.