Fluid mechanics, as all other branches in science and engineering is experiencing the data-revolutions. Thus, complex phenomena, encapsulated in some input-outputs pairs, can be modelled from the observed data, by using state-of-the-art and advanced machine learning technologies. In some cases, the learning process can be empowered by including existing knowledge within the process of learning: the thermodynamics consistency (energy conservation and entropy production), by including momentum and mass conservation, by including some choices of structural variables in complex fluids, … The use of these techniques allows enhancing, among many others: • The description of complex fluids (with complex non-Newtonian rheology) including several conformational coordinates, with theirs associated evolution equation, characterizing rich nano and microstructures (suspensions, polymers, reinforce polymers, composites, …) • Improving turbulence modeling and simulation. • Improving computational efficiency within a Model Order Reduction setting, while ensuring accuracy and stability, for addressing large-scale simulations. • Performing multi-scale simulation with the associated scales bridges. • Creating parametric surrogates. • Creating adaptive discretization techniques, empowering usual discretization techniques. • Addressing fluid-solid interaction. • Conciliating physics-based and data-driven models within a hybrid paradigm: digital twins, while assimilating data by using advanced technologies (e.g. neural-Kalman, …) • Augmenting models by discovering internal (non-observed) variables and the models governing them that impact the observed dynamics. • Taking profit of the accurate real-time responses for using that models in immersive virtual, augmented or hybrid reality. • Novel data-driven techniques for flow control