Machine learning compiler for embedded systems

Context

AI applications are also becoming increasingly widespread in embedded systems. To achieve the necessary performance and energy efficiency, dedicated accelerators are often used, e.g. Ethos-U from Arm. In order to use these effectively, specialized compilers are used to translate the trained networks into instructions that the accelerator can understand. Numerous optimizations are performed, e.g. graph and loop optimization, to increase performance.

Tasks

Your task will be to familiarize yourself with the topic of compilers for embedded systems and to adapt them for a specific accelerator. Specifically, this will involve the IREE compiler, which addresses the hardware used via a HAL. The initial focus is on the development of an integration interface.

Prerequisites

Knowledge of C++ and Python
Interest in researching AI in embedded areas
Interest in optimizations for machine learning algorithms