AI Fraud Detection

A comprehensive system designed to detect AI-generated fraud content, including deepfakes, cloned voices, and LLM-generated text.

Tech Stack

1. Programming

• Python: The core language for AI/ML development (using Jupyter for research).
• JavaScript: For frontend interactivity.

2. Frontend

• HTML, CSS, Bootstrap: For a responsive web interface.

3. Backend

• Flask: Lightweight Python framework for serving models and application logic.

4. Machine Learning

• TensorFlow / Keras & PyTorch: Primary deep learning frameworks.
• Scikit-Learn: Baseline models and evaluation metrics.

5. Computer Vision

• OpenCV: Image and video processing for deepfake detection.

6. Audio Processing

• Librosa: Audio feature extraction for voice cloning detection.

7. NLP (SMS Detection)

• NLTK: Natural Language Toolkit for text analysis and fraud detection.

8. Database

• ChromaDB: Vector database for managing embeddings and retrieval.

9. Security

• JWT Authentication & AES Encryption: Secure access and data protection.

10. Deployment & Tools

• Docker: Containerization.
• Git & GitHub: Version control and collaboration.

Directory Structure

• backend/: FastAPI application and API routes.
• frontend/: Next.js web application.
• ml_models/: ML model weights, data preprocessing, and inference scripts.