Deep Learning (DL) is a rapidly evolving methodology of applying highly non-linear parametric models to a broad range of machine learning problems. In recent years we have seen a number of very impressive practical successes of the DL approach. To name but a few, it has become the dominant approach to visual pattern recognition and the key component of modern speech recognition systems. In my talk I will introduce the audience to DL, focusing on its successful applications to Natural Language Processing (NLP). The talk will consist of three parts. The first part will cover the neural networks, the main model of deep learning, the training methods and related challenges. In the second part I will focus on DL bits and pieces that are especially relevant for NLP applications, such as word embeddings, recurrent and recursive networks. Finally, we will go through several examples of DL model solving NLP problems, such language modelling, POS-tagging, sentiment analysis, machine translation and syntactic parsing.