Artificial intelligence plays an increasingly fundamental role within any data processing system, allowing complex scenarios to be managed or particular patterns to be recognized within the data. One of the main brakes to the application of AI techniques on real systems is their low speed, due to the complexity of the algorithms used that make them unfeasible for real-time applications on systems with limited resources. IngeniArs works on the development of AI algorithms, based on neural networks, with reduced complexity, so that they can be portable on devices with limited resources or implemented within hardware accelerators, e.g. an FPGA or a VPU.