This study explores using machine learning (ML) models to optimize electric vehicle charging systems (EVCS) by addressing power grid instability and congestion. The goal is to find effective ML models for predicting power consumption and optimizing charging strategies. ML models considered include Decision Trees (DT), Random Forests (RF), Support Vector Machines (SVM), Support Vector Regression (SVR), Naive Bayes (NB), k-nearest Neighbors (KNN), Deep Neural Networks (DNN), and Long Short-Term Memory Networks (LSTM). SVR and LSTM were selected based on a literature review, with no prior comparison for EV charging in smart grids. The methodology includes data collection, cleaning, feature engineering, splitting, model training, predictions, and performance assessment using Python, Keras, and Scikit-learn with a TensorFlow backend. The dataset comprises Palo Alto EV charging sessions. Evaluation metrics MAE, MSE, and RMSE show LSTM (RMSE: 0.0415) outperforms SVR (RMSE: 0.0568).