We propose SmartEdge, an integrated architecture and system that leverages IoT devices, beyond the edge, to enable next generation deployable, compute intensive, and privacy preserving contextaware applications. SmartEdge will enable and support such applications by seamlessly leveraging idle device resources within physical proximity present beyond the edge, while also engaging relevant devices to support the intended context-aware application. To accomplish this vision, we will focus or efforts on two fundamental thrusts throughout the duration of this project. Thrust 1: Enabling accurate, minimally invasive, and robust, indoor localization that adapts to the context/need of a given application. The focus in this thrust will be on investigating solutions that are deployable in a wide range of environments that do not employ special hardware solutions. As such, we will develop solutions that utilize cellular signaling technologies with minimal cell tower associations, while leveraging inertial sensors and WiFi-based sensing as augmented modalities for obtaining higher performance in terms of accuracy and computation. We will also investigate Received Signal Strength (RSS), and Round-trip-time (RTT)-based solutions based on the new IEEE 802.11mc wireless standard. We will investigate deep learning and probabilistic approaches to improve accuracy and robustness at different levels. Thrust 2: Enabling multi-modal context aware services and applications on multiple devices. Indoor localization, is an enabler for various indoor-based applications and services. The literature today, including some of our prior work [8, 10, 76, 34], focuses on single application modalities in limited scenarios and setups. We aim in this work to leverage multiple modalities of sensing to “lock and control” a given device seamlessly without interfering with others. For example, one should be able to control an IoT device by looking at it with her head and performing hand gestures.