Developed a convolutional neural network model for image processing within the apache/singa repository, targeting the TED CT workflow. The work involved implementing multiple convolutional and pooling layers, fully connected layers, and a softmax cross-entropy loss function using Python. The model was designed with configurable training options, including distributed training and customizable optimizers, to support scalable experimentation in deep learning. Utility functions were added to streamline the creation of model instances, enabling efficient prototyping and testing. This contribution established a robust foundation for future neural network-based image analytics within the SINGA library, focusing on extensibility and practical model implementation techniques.