Amazon, by virtue of the sheer scale of its operation, stands out as one of the leading players in the digital market space. Like most competitive businesses, Amazon takes its customer and stakeholders’ reviews with great seriousness as this is the most precious feedback it gets on its products and services. Not only does our project aim to serve the objective of sentiment analysis of Amazon product reviews but its preprocessing techniques and neural network architecture also hold broader utility in natural language processing tasks. The model also presents a framework to analyse and understand user-generated content across diverse domains. This feature holds promise for contributing to advancements in customer feedback analysis, opinion mining and market research. Sentiment analysis, is one such tool that is considered critical for comprehending and exploiting massive volumes of text data created by users, consumers, and stakeholders. Our model employed tokenization, padding, and encoding techniques to preprocess the dataset. We obtained a Kaggle dataset of Amazon product reviews constructed by Xiang Zhang, pre-processed the text data and used a neural network model comprising bidirectional LSTM layers with batch normalization and dropout, achieving notable results in the classification task.