Nano-sized RF switches serve as pivotal components in current RF applications, facilitating critical aspects such as miniaturization, low power consumption, minimum pull-in voltage, high-frequency operation and enhanced RF performance. The fixed–fixed beam-based Radio Frequency (RF) Nano Electro-Mechanical System (NEMS) has been chosen and designed in COMSOL Multiphysics software to analyze its electrical characteristics. Minimum pull-in voltage (Vp) is the utmost need for such nano-sized switches which is obtained by optimization of the dimensional parameters of the switch. Such optimization necessitates a time-consuming trial-and-error process. To streamline the design process, this paper introduces an Artificial Neural Network (ANN) model aimed at predicting the Vp from nano switch dimensions. Employing various training algorithms, the ANN model achieves enhanced performance and computational efficiency. Predicted Vp from the Levenberg–Marquardt backpropagation neural network model closely align with simulation software values and findings from previous experimental studies. The developed LM based neural network model accurately predicts a minimum pull-in voltage of 2.51 V, demonstrating superior performance with a reduced Mean Square Error (MSE) of 0.000092. The optimized physical dimensions of the NEMS switch with minimum Vp is designed to observe the RF performance and it is found that it gives desirable S parameters in the frequency range of 1–10 GHz.