Developed a Deep Learning-based Network Intrusion Detection System (NIDS) using a Bidirectional LSTM Encoder and a Context-Conditioned WGAN-GP architecture for unsupervised anomaly detection. The system was trained exclusively on normal network traffic from the NSL-KDD dataset to detect zero-day attacks and previously unseen cyber threats. Built a complete preprocessing pipeline including feature encoding, scaling, and sliding-window sequence generation. Implemented a hybrid anomaly scoring mechanism combining reconstruction error, discriminator confidence, and feature deviation. Evaluated the model using Precision, Recall, F1-Score, ROC-AUC, and Confusion Matrix. Also developed an interactive CustomTkinter desktop dashboard for real-time monitoring, alerting, and XAI-based feature importance visualization.
