This project aims to create a way to enable communication between two frameworks. Such as allowing a simulation in say Unity, to continuously send packets of data, that can be used by another computer in the local network to learn, or make inferences and return a suitable response. This can be useful for computationally expensive tasks, where both the simulation and the neural network are typically huge. In the link, you will see a blog indicating a rudimentary implementation of a local server client connection which has a simple neural network.
Looking for collaboration to improve the speed, and alternate methods to do this in a local network or even over the internet.