Over the last years, hybrid genetic search (HGS) algorithms for the VRP and its variants have demonstrated encouraging results in particular with the use of a route-first-cluster-second heuristic. As an individual is represented by a “giant tour” (TSP Solution) in this method, the advantages are an easiness regarding the population management and crossover operators combined with an optimal split method for the transition from an individual to a VRP solution. In parallel progress in optimization using deep learning model have been significants, especially for the vehicle routing problem (VRP). The goal is to improve the efficiency of HGS by taking advantage of deep learning methods to evaluate cost directly on an individual, thus avoiding the need to switch search space and saving computational time. To this end, a reverse split method is used to switch freely between population and solution space to train the deep learning model using cost evaluation and feature extraction.