This paper introduces an innovative approach to the pressing conservation challenge of accurately identifying endangered bird species, with a focus on Nepal's diverse avian population. Using Convolutional Neural Networks (CNNs), our deep learning system achieves impressive accuracy in classifying bird species from images. We compile a comprehensive dataset of 8,457 high-quality images representing 38 endangered bird species native to Nepal, sourced from various online platforms. Through meticulous data augmentation, we enhance dataset diversity and model robustness. Four CNN models are developed and rigorously evaluated, with test accuracies ranging from 83.29% to an impressive 90.8%. The highest-performing model is seamlessly integrated into a user-friendly web application built on Django, allowing users to upload bird images for real-time classification. Our findings highlight the potential of deep learning in advancing conservation efforts, offering scalable solutions for monitoring and protecting endangered avian populations. This work contributes to the intersection of artificial intelligence and conservation biology, demonstrating the crucial role of technology in preserving Earth's biodiversity.