A broad class of important engineering applications, including intelligent transportation systems, smart grids, and smart factories can significantly benefit from novel methodological approaches that: (i) involve cooperation among agents with local computation and communication capabilities (distributed computing) and (ii) exploit novel Machine Learning approaches to enhance the autonomy of the complex system. This novel paradigm, combining distributed computing and Artificial Intelligence (AI), calls for novel numerical methods that lie at the intersection of optimization theory, network systems, and Machine Learning. While most of the available optimization-based methods in Machine Learning are inherently centralized, in the above-mentioned scenarios a distributed optimization paradigm is needed due to the presence of big-data problems in which data are spatially distributed and private. The main objective of the MAECI project is to model distributed learning and control problems in cyber-physical networks as distributed optimization problems, solve them with new distributed algorithms, and then apply them to scenarios from domains such as data analytics, traffic control, and cooperative robotics.