In the contemporary field of information, combating Fake information on social media has become a formidable challenge. The Research paper employs on machine learning strategies, particularly focused on feature engineering, to enhance the accuracy of fake news detection. This involves selecting, transforming, and extracting relevant attributes from raw data to provide discriminative information to machine learning models. The pre-processing stage involves crucial steps such as identifying textual features, where the analysis of language patterns and sentiment helps discern Legitimate from Fake News. The methodology framework integrates ensemble machine learning models, such as Support Vector Machines, Random Forests, Artificial Neural Networks and Convolutional Neural Networks to effectively predict the authenticity of news articles or social media posts. The research explores an overall architecture that combines feature engineering models with neural networks for a comprehensive set of features. Experimental results show that ensemble classifiers, particularly combinations like Random Forest Classifier with Support Vector Machine (97.01), Artificial Neural Network (97.23) and with Convolutional Neural Network (98.21) significantly outperform individual classifiers. However, challenges such as dataset overfitting underscore the need for continual research and innovative approaches to address the evolving landscape of fake news on public platforms.