Machine-learning (ML) algorithms are increasingly finding their way into the field of fluid mechanics. Building on the remarkable successes of these methods in various scientific and engineering applications, this research explored the application of linear genetic programming (LGP) to control jet mixing using a coaxial dielectric barrier discharge actuator (DBD) positioned at the nozzle exit. In a continuous actuation mode, the plasma actuator induces secondary flow that can affect the velocity gradient of the free shear layer near the jet nozzle. The effectiveness of mixing is assessed by monitoring the centerline mean velocity, measured using a hotwire sensor positioned at a distance of 5D, which marks the end of the potential core. Increased entrainment near the potential core is correlated with a reduction in centerline velocity, contributing to improved mixing of the primary jet flows. To enhance the mixing, this study investigates the impact of varying the duty cycle and burst frequency \(f_{b}\) of the plasma actuator on the primary flow. The study identifies the optimal frequency ratio \(f_{b} /f_{{\text{o}}}\) as 1.7, where \(f_{0} \) represents the preferred mode frequency of the main jet. Moreover, instead of keeping burst frequency and duty cycle (DC) constant, these parameters are dynamically adjusted to promote mixing further. For this purpose, a model-free strategy called “Machine Learning Control (MLC)” is employed in control design. MLC relies on an evolutionary algorithm that modifies control laws until the cost function is minimized. Remarkably, MLC identifies an optimal control law that matches the performance of the optimal burst actuation with only 300 actuation measurements, each test conducted for 5 s. Subsequently, Schlieren visualization is performed to verify the effectiveness of the control strategy.