Active Learning (AL) is a subfield of machine learning where the model can actively query a human or oracle for labels on selected data points, thereby optimizing its learning process. Active learning is particularly useful in scenarios where labeled data is scarce or expensive to obtain, and the goal is to achieve high model performance with minimal annotation effort.Active Learning plays a crucial role in reducing annotation costs while improving model performance, especially in resource-constrained settings where labeled data is expensive or difficult to obtain. The above PhD project ideas span a wide range of domains, from medical imaging and autonomous driving to natural language processing and reinforcement learning. These projects offer exciting opportunities to explore the intersection of deep learning, data efficiency, and human-in-the-loop systems, advancing the state of the art in active learning research.