There's been a lot of interest lately on large language models, with an emphasis on the large. Even tinybox is targeting multiple GPUs with a cost over $10k.

But ML can also by truly tiny, running on embedded systems including smartphones, single-board computers, and microcontrollers. These systems typically target real-time ML applications. Typically, this means a batch size of one and consuming streaming sensor data as input. This use case, combined with the unique constraints of embedded systems on on power, size, and computational resources, present unique challenges and opportunities for deploying these models

In this blog, we'll share insights, tutorials, case studies, and industry news related to real-time ML in embedded systems. Whether you're an experienced ML practitioner looking to expand your knowledge or a hobbyist exploring the intersection of these technologies, we hope you will find our content informative and inspiring.