This research proposal aims to analyze the capabilities of Machine Learning to reconstruct the input features of an aberrated wavefront transmitted through an optical fiber, in view of focal plane wavefront sensing applications in ground-based telescopes. Future astronomical instrumentation, in fact, is increasingly moving towards photonic waveguides to carry information and fulfill stringent requirements for data acquisition and sensing. Recent studies have provided insight into the generation of phase distortion on light propagation through fibers. Furthermore, Machne Learning based reconstruction methods, as applied to multimode and multicore fibers, have already demonstrated their potential in specialized medical fields. Building upon this knowledge, our objective is to study the feasibility of employing similar techniques for the development of astronomical instrumentation, thereby expanding the horizons of astrophotonics applications