Now it does not need a complicated hardware to design your own TensorFlow model. The following simple and very tiny version of kit, TensorFlow Lite from Adafruit can perform Machine Learning computations.
This kit uses Adafruit PyBadge board (SAMD51 Cortex M4F prcoessor@120MHz) as your edge processor. Its size is about the size of a credit card. It’s powered by ATSAMD51, with 512KB of flash memory and 192KB of RAM. It also features 2 MB of QSPI flash memory for file storage, making it easy to store TensorFlow Lite files, images, fonts, sounds, or other assets.
If desired, it can plug a microphone into the port on the bottom to add a microphone input for micro-speech recognition. TensorFlow Lite for microcontrollers is very cutting-edge, so expect a lot of development in this area, including a lot of code and process changes.
To learn more TensorFlow Lite, please take this quickstart