This work presents an artificial neural network (ANN)-tuned proportional integral (PI) controller design algorithm to control the water level of nonlinear systems. Conventional PI controllers are working well in linear systems; it degrades the performance of nonlinear systems in the presence of nonlinearity. The tuning methods are complex in a highly nonlinear system. This work proposes an ANN-tuned PI controller as a solution to these problems. The ANN-tuned PI controller algorithm is built based on a time series nonlinear exogenous (NARX) model, a linear layer neural network model (INST), and a general minimum variance (GMV). For nonlinear systems, the NARX model predicts the one step ahead output, and the GMV model minimizes error and determines the best combination of PI gains to reduce the steady-state error. The proposed controller algorithm is simulated to control water level in the conical and spherical tank systems. To show the efficacy of the designed controller, comparative analysis is done with particle swarm optimization (PSO) algorithm-tuned PI controller in terms of integral square error (ISE), integral absolute error (IAE), and time domain specification. Furthermore, real-time validation of the proposed method is performed on the interacting two-tank experimental setup and the performance of an artificial neural network (ANN) optimized PI controller has been examined, revealing its superiority over alternative control mechanisms in both transient and steady-state responses of nonlinear processes.