This paper presents a deep learning-based, non-invasive blood group identification system using fingerprint biometrics. The suggested method preprocesses high-resolution fingerprint photos and uses a Convolutional Neural Network (CNN) architecture tuned using data augmentation approaches to improve classification accuracy. The model achieves 87.5% accuracy in training and 85.5% in validation, effectively capturing intricate fingerprint patterns associated with specific blood types. Experimental results across multiple blood groups demonstrate strong performance in terms of accuracy, F1-scores and recall. By eliminating the need for invasive blood sampling, this AI-driven method introduces a novel biometric alternative for blood group prediction, supporting wider accessibility in healthcare. The work corresponds with SDG 9 (Industry, Innovation, and Infrastructure) and SDG 3 (Good Health and Well-Being).