In this paper, we propose an energy-efficient radar beampattern design framework for a Millimeter Wave (mmWave) massive multi-input multi-output (mMIMO) system, equipped with a hybrid analog-digital (HAD) beamforming structure. Aiming to reduce the power consumption and hardware cost of the mMIMO system, we employ a machine learning approach to synthesize the probing beampattern based on a small number of RF chains and antennas. By leveraging a combination of softmax neural networks, the proposed solution is able to achieve a desirable beampattern with high accuracy.