In the growing advancement of Natural Language Processing, sarcasm detection, a part of sentiment analysis is still a challenge. It is important to detect sarcasm in the statement, especially in open platforms where people are free to express themselves, to identify the correct intended meaning of the statement mentioned. Sarcasm is an important processing problem in the NLP field of sentiment analysis, as it serves as an interface between communicating humans and machines, also such sentences can change the polarity of a sentence, which might result in wrong sentiment analysis. Here, in this project, we going to work upon sarcasm detection in news headlines, which we can get online or in news articles. We made use of LSTM[1][4], an artificial recurrent neural network architecture used in the field of deep learning[8], built around our collected dataset. The model is trained over the pre-processed dataset, cleaned with the help of the nltk library enabling it to create word embeddings, remove stopwords, etc.