Neural machine translation is a recently introduced approach to tackling machine translation. Unlike the conventional approach, neural machine translation aims to build a translation system that consists of a single neural network, where the whole system is jointly trained to maximize the conditional probability of a translation given a source sentence. In this talk, I will first introduce a basic concept of neural machine translation and demonstrate the working neural translation system. Then, I will describe how we incorporated attention mechanism into an encoder-decoder model to achieve translation performance comparable to the conventional phrase-based machine translation system (http://arxiv.org/abs/1409.0473)