This research aims to improve cybersecurity measures by addressing cyber threats and promptly identifying adverse events and anomalies in real-time. A primary focus is the development of a novel Intrusion Detection System (IDS) that overcomes the limitations of existing state-of-the-art systems. This research applies state-of-the-art approaches of CV,NLP domains like BERT to develop robust classification of network traffics. This innovative approach bridges the gap between conventional cybersecurity practices and foundational models, fostering the emergence of more robust IDSs. The study currently investigates pretrained model for IDS models and compare their deployability to hardware for real-time detection.