A crucial component of network security is intrusion detection, but this task can be difficult when dealing with statistics on the distribution of network traffic, where benign cases exceed malicious ones by a large margin. The goal of this project is to create a powerful intrusion detection system using a recurrent neural network (RNN) to address this problem. RNNs are excellent at identifying temporal dependencies in sequential data, which makes them useful for examining network traffic patterns. The suggested methodology calls for preprocessing the data, balancing the dataset, feature extraction, choosing an RNN model, training, assessing, adjusting, and monitoring in real time. The network traffic data is cleaned and formatted during the preprocessing stage, and balancing algorithms make sure the RNN is trained on a representative dataset. It then extracts pertinent aspects, including behavioral patterns and packet header data. Effective intrusion detection requires the use of an appropriate RNN architecture, such as LSTM or GRU. The model is subsequently trained, assessed using a range of criteria, and refined through iterations. The optimized RNN model is then implemented for real-time intrusion detection, monitoring network traffic continuously and producing alerts for probable intrusions. This research seeks to improve the precision and efficacy of network security systems, maintaining the integrity of crucial information in the linked world of today, by utilizing RNNs for intrusion detection in imbalanced network trafficking. Keywords IDS, Imbalanced Network Traffic, Deep Learning, Recurrent Neural Network.