In this talk, I will showcase some of the recent advances in deep learning that has fundamentally changed several research directions such as computer vision and natural language processing. A brief overview will be given on deep convolutional and recurrent networks, the types of problems they are solving in computer vision and natural language processing, and intuitions about why they work. A brief demo will be given to showcase how to easily build a deep neural network. I will also briefly talk about our recent research in adapting convolutional neural networks to non-standard domains such as prediction in spectrograms, as well as helping human understanding of the deep networks.