AI Security Sentinel

Overview

AI Security Sentinel is a modern, microservices-based Web Application Firewall (WAF) designed to detect and intercept malicious web traffic in real-time. Built with a powerful AI engine, it accurately identifies SQL Injection (SQLi) and Cross-Site Scripting (XSS) attacks before they reach the core application, providing an intelligent layer of defense beyond traditional rule-based systems.

Key Features

• Deep Learning Detection: Utilizes a Bidirectional LSTM (Bi-LSTM) model trained with TensorFlow to analyze character-level payload sequences and catch obfuscated attacks.
• Real-Time Traffic Interception: A NestJS backend equipped with a custom security interceptor to evaluate incoming HTTP requests and block threats instantly.
• High-Performance Microservice: The AI model is deployed as a standalone, lightweight FastAPI endpoint to ensure minimal latency during request screening.
• Interactive Monitoring Dashboard: A responsive Next.js frontend utilizing Recharts to visualize network traffic, safe requests, and blocked threats in real-time.

Challenges & Solutions

Challenge: Traditional rule-based WAFs struggle with heavily obfuscated payloads, leading to false positives or bypassed security. Furthermore, analyzing every single HTTP request with an AI model can introduce significant latency, degrading the user experience.

Solution: I designed a hybrid detection pipeline. It first passes through a lightning-fast rule-based filter for obvious threats, and only suspicious payloads are sent to the AI engine. To tackle latency, the architecture was split into microservices, offloading the heavy computational AI tasks to an optimized FastAPI server. This ensures robust, deep-level security analysis while maintaining high system throughput and low response times.