We contribute a thin, transparent, and low-cost design for electric field sensing, allowing for 3D finger and hand tracking and gestures on mobile devices. Our approach requires no direct instrumentation of the hand or body, and is non-optical, allowing for a compact form-factor that is resilient to ambient illumination. Our simple driver electronics are based on an off-the-shelf chip that removes the need for building custom analog electronics. We describe the design of our transparent electrode array, and present a machine learning algorithm for mapping from signal measurements at the receivers to 3D positions. We demonstrate non-contact motion gestures, and precise 3D hand and finger localization. We conclude by discussing limitations and future work.