A significant portion of the material market is completely governed by the novel polymer composites, either reinforced with synthetic or natural fibers. However, machining the fiber-reinforced composite (FRP) poses a significant challenge, particularly in terms of kerf and delamination damage during the abrasive water jet machining (AWJM). This study advocates the novel development of a banana fiber woven-pumice particle reinforced sandwich composite and further machining quality improvement during AWJM by implementing a hybrid statistical model, Grey Relational Analysis (GRA)-Principal Component Analysis (PCA), and soft computing Machine Learning (ML) algorithms. Nine machining trials are executed by considering machining parameters like pumice composition, hydraulic pressure, abrasive flow rate, and standoff distance, where kerf ratio, delamination, and surface roughness are the machining damages. Invariably, the models considered declare the ideal set of parameters, which comprises 100 (MPa) hydraulic pressure, 300 (gm/min) abrasive flow rate, 2 mm standoff distance, and 10% pumice particles to ensure minimal damage. Both modeling approaches indicate that the pumice composition in the sandwich core is the highest contributor in deciding the machining quality. The implementation of the hybrid model facilitates the 7.88%, 24.43%, and 45.28% improvement in damage reduction in kerf ratio, surface roughness, and delamination.