Bi-directional power flow made possible by vehicle-to-grid (V2G) technology in grid-interactive electric vehicle (EV) chargers lets EVs not only consume but also provide electricity from the grid. This capacity improves grid stability and energy management and provides financial gains, such as selling electricity during peak demand. Although advancements in power electronics and converter topologies have increased system efficiency, it is still difficult to ensure power quality under less-than-ideal grid conditions. The study explores the use of machine learning (ML) in conjunction with the decision tree learning approach to replace traditional PI controllers, therefore reducing total harmonic distortion (THD) and improving grid stability. Finally, a 1.1 kW MATLAB/Simulink simulation supports the efficiency of the proposed system.