The Internet of Things (IoT) will connect billions of devices to the Internet and create a large-scale dynamic and open environment with high heterogeneity. To assure rapid adoption of IoT applications, application developers and users need to be abstracted from IoT infrastructure via scalable middleware. Event-processing systems have the potential to contribute in filling the gap between the IoT infrastructure and applications layers. Event processing follows a decoupled model of interaction in space, time, and synchronization. However, the dimension of semantic coupling still exists and poses a challenge to scalability in highly semantically heterogeneous and dynamic environments such as the IoT. Here, the authors describe an approach based on loosely coupled producers and consumers enabled with approximate semantic matching of events. They emphasize a practitioner perspective to IoT architectures for building software that can tackle heterogeneity of IoT events.