Building Resilient Machine Learning Pipelines with IoT Data

Keewi’s machine learning pipeline is built on our network of connected IoT infrastructure of smart plugs and hubs. Like any other physical system, IoT networks can behave in surprising ways, which present unique challenges for downstream ML development. IoT oddities can stem from user behavior, network connectivity changes, and the accuracy of the sensors themselves. We’ll discuss tools that Keewi uses to minimize expected challenges and our strategies in preparing for unexpected ones.