🛡️ SentinEL: Ultima Intelligence Engine

Next-Gen Phishing Detection using Explainable AI (XAI)

Architect: Mourya R. Udumula | Ops Lead: Jeet Upadhyaya Institution: Indrashil University

---

🚀 Project Overview

Technical Stack

• Intelligence Layer: Scikit-Learn (Random Forest Pipelines), Feature Scaling (Standardization).
• SentinEL Ultima is a Hybrid Threat Intelligence Engine designed to detect sophisticated phishing attacks that bypass traditional blacklists. It combines a high-speed Random Forest classifier with real-time Forensic Analysis (WHOIS, DNS, SSL) to deliver verdicts withForensic Modules: python-whois, dnspython, Socket-level SSL/TLS Handshake inspection. <150ms latency.

The system features Active Learning, allowing security analysts to flag false positives and ret* Computational Logic: Optimized Shannon Entropy algorithms to quantify string randomness.

• Interface: Streamlit-rain the decision boundary in real-time (Session Scope).

🛠️ Technical Stack

• **MLbased Dashboard for real-time local XAI Attribution.

⚡ Key Capabilities

1. Hybrid Engine: Seamlessly integrates Allowlisting (O(1) lookup) with ML Heuristics.
2. Explainable AI (XAI): Human-readable forensic justifications (e.g., "High Entropy", "Expired SSL").
3. Adversarial Resilience: Detects DGA (Domain Generation Algorithms) via algorithmic entropy analysis.
4. Active Learning Feedback: Real-time analyst-override mechanism (Session Scope) to refine decision boundaries.

🔧 Installation & Usage

# Clone the intelligence engine
git clone https://github.com/Maze-6/SentinEL-Adversarial-ML.git

# Install dependencies
pip install -r requirements.txt

# Launch the engine
streamlit run app.py