The ultimate goal of materials science is to be able to adapt microstructures to achieve desired properties. However, to date, no consistent constitutive models have been developed due to the need to statistically link the microscopic and macroscopic scales. In this project, we propose an original methodology where a crystalline plasticity code will be coupled with a supervised learning algorithm to create a system capable of suggesting the distribution of operating mechanisms in a polycrystal with its microstructural parameters in order to achieve desired macroscopic mechanical properties. This new model resulting from the learning process will be trained using a large set of experimental data obtained through scanning electron microscopy and translated into an input-output system. This project will have a significant impact on current societal issues by enabling energy savings and reducing costs associated with the tuning of microstructures targeting specific mechanical performances.