Graph Representation Learning (GRL) is a rapidly advancing field in machine learning that focuses on learning low-dimensional, informative embeddings of graph-structured data, enabling efficient analysis and prediction tasks. Unlike traditional handcrafted graph features, GRL leverages neural architectures to capture node attributes, edge relations, and global graph topology. Early research explored random-walk-based methods such as DeepWalk, node2vec, and LINE, while recent advances center on graph neural networks (GNNs), graph convolutional networks (GCNs), graph attention networks (GATs), and message-passing neural networks (MPNNs). These models have been applied across domains including social network analysis, recommender systems, bioinformatics, chemistry (molecular property prediction, drug discovery), cybersecurity, and knowledge graphs. Current research explores dynamic and temporal graph representation learning, heterogeneous graphs, scalable training for large graphs, and integration with self-supervised learning, contrastive learning, and transformers for graphs. Additionally, robustness, interpretability, fairness, and privacy-preserving graph learning are emerging focus areas. Collectively, GRL is a cornerstone for advancing intelligent systems that require reasoning over relational and interconnected data.