Project Aegis: AI Cyber Defense Assistant 🛡️
An AI-powered SOC teammate that detects network anomalies, clusters threat behaviors, and translates raw telemetry into actionable, plain-English incident summaries.
🚀 Overview
Modern Security Operations Centers (SOCs) are overwhelmed by alert fatigue. Project Aegis shifts the paradigm from manual log analysis to AI-driven proactive defense. By combining high-speed machine learning for threat detection, Explainable AI (XAI) for transparency, and Large Language Models (LLMs) for reasoning, Aegis acts as a Tier-1 SOC Analyst that never sleeps.
✨ Key Features
- Real-Time Threat Engine: High-throughput network anomaly detection powered by an optimized XGBoost classifier.
- Explainable AI (XAI): Sub-millisecond SHAP (SHapley Additive exPlanations) integration that pinpoints the exact network features triggering an alert, ensuring analysts trust the AI.
- Autonomous Incident SITREPs: Integration with Google Gemini to automatically translate raw numerical telemetry and SHAP values into concise, plain-English Situation Reports with actionable remediation steps.
- Interactive SOC Dashboard: A responsive, live-telemetry UI built with Streamlit for seamless analyst interaction.
🛠️ Tech Stack
- Core AI/ML: Python, XGBoost, Scikit-Learn, Pandas, NumPy
- Explainability: SHAP
- Generative AI: Google Gemini (gemini-2.5-flash)
- Frontend UI: Streamlit
🗄️ Dataset
This project was designed around the UNSW-NB15 dataset, a comprehensive network intrusion detection dataset containing normal and malicious network traffic across multiple attack categories (Fuzzers, Backdoors, DoS, Exploits, etc.).
(Note: The provided app includes a dynamic fallback generator for immediate testing if the raw dataset is not present locally).
⚙️ Quick Start Guide
Prerequisites
- Python 3.9+
- A Google Gemini API Key
Installation & Execution
- Clone the repository:
```bash
git clone https://github.com/426024-rgb/Project-Aegis-AI-Cyber-Defense-Assistant.git
cd Project-Aegis-AI-Cyber-Defense-Assistant