The widespread adoption of wearable sensors in sport science and health monitoring has exposed the need for data-processing infrastructures that are simultaneously reliable, extensible, and reproducible. Such infrastructures must support a coordinated sequence of operations: receiving data streams from heterogeneous sources; pre-processing and time-synchronising those streams; filtering and cleaning noisy or incomplete readings; and executing processing algorithms to extract meaningful information. These requirements motivate a flexible and modular infrastructure, where each component can be adapted, replaced, or extended without redesigning the entire system. This thesis presents the design, implementation, and validation of Wearable IoT Activity Pipeline, a containerised Internet-of-Things (IoT) platform for wearable sensor acquisition, real-time data streaming, AI-based activity classification, timeseries storage, and heterogeneous sensor integration.