Deep learning algorithms seek to exploit the unknown structure in the input distribution to discover unique representations, often at multiple levels, with higher-level learned features defined in terms of lower-level features. Deep learning has vastly improved our ability to understand and analyze image, sound and video and has quickly become the underpinning of many advanced machine learning applications today. In this presentation we present the concepts of implementing a Deep Learning platform. The subjects we describe include: (i) the architecture and key components of the platform; (ii) the technical steps for training and building various deep-learning networks such as deep forward networks (DNNs), convolutional neural networks (CNNs), and recurrent neural networks (RNNs); (iii) applications to supervised learning tasks like computer vision and natural language processing; and (iv) extensions to unsupervised and transfer learning. Using compelling case studies, relatable scenarios and lessons learned, we will also discuss about practical aspects of deploying deep learning models and model lifecycle management in real-world environment.