The advent of Artificial Intelligence (AI) in agriculture has sparked a paradigm shift toward data-driven smart farming. This paper examines the role of machine learning (ML) in precision agriculture, presenting a framework for crop yield optimization using real-time and historical agricultural data. The study outlines the research problem of inefficient resource use and low productivity in traditional farming, motivated by the need to ensure food security for a growing global population. Through a review of supervised, unsupervised, and deep learning applications, the methodology integrates satellite imagery, sensor data, and intelligent models to forecast yields and automate decision-making. Findings highlight improved efficiency, with recommendations for policy, scalability, and farmer inclusion.