The optimization of Time-Synchronized Channel Hopping (TSCH) scheduling at the edge plays a critical role in enhancing the efficiency and reliability of the Internet of Things (IoT). In this paper, we propose an approach leveraging Deep Neural Networks (DNNs) for TSCH scheduling. Our approach aims to harness the capabilities of hardware accelerators such as the MAX78000 to achieve efficient and reliable scheduling. We train a DNN using a database generated with the Traffic Aware Scheduling Algorithm (TASA) and evaluate its performance in generating schedules, achieving an accuracy of over 94%. Our experiments demonstrate significant reductions in execution time, with the DNN-based approach achieving an average reduction of 99.69% compared to running TASA on the same microcontroller. Similarly, a substantial reduction in energy consumption is observed, with the DNN-based approach consuming only 5uJ compared to 1,606uJ by TASA running in software. This represents a significant milestone in the field low-power wireless networks, as it opens up new possibilities for implementing sophisticated AI-based solutions directly on edge devices, without the need for extensive computational resources.