The high rate of the development of digital technologies has complicated the forensic investigations and requires intelligent and automated systems to conduct the effective crime analysis. In this paper, the author introduces Cipher Eye - an AI based Forensic Investigation Framework that combines several deep learning modules to assist the law enforcement in analysis of evidence and identification of suspects. Extensive experiments on benchmark datasets, such as CASIA2.0, show that Cipher Eye achieves competitive performance and better interpretability comparing with state-of-the-art methods. It has been proposed to include four key components in the proposed system: (i) a forensic sketch generation system that generates facial sketches based on CCTV images or eyewitness accounts and can identify nearly 90% similarity between them, (ii) a tampered image detector system that evaluates the authenticity of visual evidence with the help of CNN-SVM hybrid deep learning models, with 94% accuracy, (iii) a face recognition system that identifies the faces of suspects in a criminal database with the help of the FaceNet, with accuracy between 95%–97% The model-level analysis of the system based on a confusion matrix and classification report show all high precision, recall, and F1-scores, which proves the effectiveness of the deep learning architecture. The integrated framework is implemented such that it is a web-based application and it can be used in real-time to analyze and visualize it to be used practically by forensic experts. The study also shows that various modules of AI can be integrated to increase accuracy, reliability, and efficiency in the digital and criminal forensic investigations.