The increasing demand for electricity and the integration of smart grid technologies have highlighted the need for advanced energy monitoring solutions. Non-Intrusive Load Monitoring (NILM) is essential for breaking down total energy consumption into individual appliance-level data, enabling more efficient and sustainable energy management. This work introduces the development and deployment of an IoT-enabled NILM system tailored for residential spaces, leveraging deep learning techniques to enhance classification accuracy. The system integrates a smart meter for real-time event detection and utilizes WaveNet, a deep neural network originally designed for speech processing, to classify appliance ON/OFF events based on electrical parameters. A comparative analysis with 1D-Convolutional Neural Networks (1D-CNN) and Long Short-Term Memory (LSTM) demonstrates that the WaveNet model can also achieve high classification accuracy, reaching 98.84%. The system’s performance was validated using real-world residential data, showcasing its practicality and scalability for real-time energy monitoring. These findings contribute to advancing NILM research by demonstrating the potential of deep learning models in smart energy applications.