This project represents a significant advancement in web application firewall (WAF) technology by integrating real-time prediction with advanced deep learning models, specifically long short-term memory (LSTM) and recurrent neural networks (RNN). These models are selected for their effectiveness in handling sequential and time-sensitive data, with a primary focus on identifying and thwarting malicious web requests in real time, particularly targeting common attacks such as SQL injection and cross-site scripting. Utilizing the NSL-KDD dataset, a widely recognized benchmark for network intrusion detection, comprehensive training, and assessment of the models were conducted. The results demonstrate robust pattern recognition capabilities throughout both training and validation phases, highlighting their capacity to learn and accurately anticipate potential web attack threats. This project's endeavor to apply real-time prediction to web application firewall technology signifies a significant advancement in developing proactive and resilient cybersecurity defenses. The integration of these cutting edge techniques promises enhanced protection against evolving cyberthreats, contributing to a safer digital landscape.