Unsupervised Representation Learning is a critical area of machine learning that aims to automatically extract meaningful features from unlabeled data. This process is essential for tasks where labeled data is scarce or expensive to acquire. Advances in unsupervised learning, particularly with deep neural networks, have led to significant breakthroughs in areas such as natural language processing (NLP), computer vision, and reinforcement learning.Unsupervised Representation Learning offer exciting opportunities to advance the field of machine learning by focusing on real-world challenges where labeled data is scarce. These projects cover a variety of domains, including anomaly detection, visual representation learning, graph data, NLP, time-series data, reinforcement learning, and healthcare. Each project explores novel ways of extracting meaningful representations from data, ultimately improving performance across a wide range of downstream tasks.