Features multiple integration of heterogeneous AI platform/system and ASIC-based module. For example, an embedded ARM platform or X86 system is combined with AI accelerators such as M.2 and MXM. In general, the series highlights low latency, low memory usage, whereas high degree of customization services for accelerating neural network (NN) deployment and deep learning (DL) process according to the specific AI applications.