Summary. The popularity of smartphones equipped with GPS and geomagnetic sensors has spurred mobile application developer interest in augmented reality (AR), which presents highly contexualized, spatially relevant information that enhances user knowledge of their immediate surroundings. AR applications typically mesh relevant information with user views of the physical world. Prior research has focused on interfaces built with custom hardware, but a smartphone equipped with GPS, a camera, and a geomagnetic sensor is an attractive alternative to traditional solutions. These devices can be programmed to present context-sensitive information to users without needing custom hardware. This paper examines three key challenges facing AR developers on mobile devices and presents solutions applicable to modern mobile platforms, such as Apple's iPhone and Google Androidbased smartphones. First, we investigate methods of filtering raw sensor data and present an algorithm that eliminates sensor noise. Second, we explore the process of implementing a "magic lens" interaction metaphor by overlaying perspective-rendered graphics on the device's camera using OpenGL and UIKit. Third, we provide an efficient technique for fetching and caching geographically tagged points of interest from a server.