Sentiment Analysis Using Deep Learning Is The Process Of Automatically Identifying And Categorizing Emotions Expressed In Textual Data By Leveraging Neutral Network Models That Learn Complex Pattern In Language. This Paper Presents A Robust Framework For Sentiment Classification Of Social Media Text Using A Bidirectional Long Short – Term Memory (BiLSTM) Model . A BiLSTM Based Model Is Use To Analyse Social Media Data, With Preprocessing, Class Balancing, And Feature Extraction Stages. The Model Achieves 95% Accuracy , Effectively Predicting Positive, Negative , And Neutral Sentiments , Demonstrating Its Strength For Real-Time Sentiment Monitoring