Deep Neural Networks are considered as one of the most powerful machine learning methods of recent times. Recurrent neural networks, including LSTM variations, exhibit exceptional performance in sequence-oriented assignments, while also falling within the domain of autoregressive models, wherein forecasts are tied to the historical input context. In this paper, we experiment with LSTM for Twitter sentiment analysis. Leveraging advances in Natural Language Processing (NLP), we show the efficacy of our algorithm with extremely competitive results.