🛡️ TruthLens AI

AI-powered trust and safety platform for detecting suspicious messages, phishing patterns, risky URLs, and potentially dangerous online content.

Problem

People receive suspicious emails, SMS messages, social-media messages, verification requests, payment requests, and URLs every day. A simple fake/real label does not explain why content is risky or what the user should do next.

TruthLens AI combines machine-learning classification with security signals to produce an explainable, risk-oriented result.

Key Features

• 🤖 Word + character TF-IDF text classification
• 🧠 ML probability and risk scoring
• 🚨 Security signals for urgency, financial language, credential requests, and action requests
• 🔗 URL analysis for suspicious structural characteristics
• 🔍 Reasons and recommended actions
• 📚 Analysis history
• 🌐 FastAPI backend + React frontend

Architecture

User
 │
 ▼
React Frontend
 │
 ▼
FastAPI API
 │
 ├──────────────┬──────────────┐
 ▼              ▼              ▼
ML Classifier   Risk Engine    URL Analyzer
 │              │              │
 └──────────────┼──────────────┘
                ▼
        Explainable Result
                │
                ▼
          SQLite History

ML Performance

The documented V2 evaluation uses a leakage-aware validation setup.

Metric	Result
Accuracy	98.81%
Precision	98.85%
Recall	98.78%
F1 Score	98.82%
ROC-AUC	0.9989

These are project evaluation results, not a guarantee of real-world detection performance.

Tech Stack

Layer	Technology
Frontend	React
Backend	Python + FastAPI
ML	scikit-learn
NLP	Word + character TF-IDF
Database	SQLite
Tooling	Git + GitHub

Workflow

Message / URL
     ↓
Preprocessing
     ↓
ML Classification + Security Signals
     ↓
Risk Scoring
     ↓
Explainable Analysis
     ↓
Recommended Action
     ↓
History / Dashboard

Project Status

Active portfolio project. The core ML analysis, risk-oriented workflow, URL signals, and full-stack structure are implemented. The next stage focuses on automated testing, authentication, production data storage, containerization, CI/CD, monitoring, and cloud deployment.

Roadmap

• Leakage-aware model evaluation
• Word + character TF-IDF model
• Risk-oriented analysis
• URL signal analysis
• FastAPI backend
• React frontend
• Analysis history
• Automated test suite expansion
• Authentication
• Production database
• Docker deployment
• CI/CD
• Model monitoring
• Cloud deployment

Author

Sahibzada Aizaz Ur Rahman
Python Developer | AI/ML Engineer

• GitHub: https://github.com/aizaz512
• Portfolio: https://github.com/aizaz512/sahibzada-portfolio

⭐ If you find TruthLens AI useful, consider starring the repository.