Spandan-Satapute/ai-email-security-system

🚀 AI Email Security System

An intelligent AI-powered system that detects spam emails and evaluates their risk levels using machine learning. Built as a real-world OpenEnv environment, where an agent learns step-by-step through tasks.

🔥 Live Demo

👉 https://huggingface.co/spaces/Spandan-Satapute/ai-email-security-system-v2

🧠 Problem Statement

Spam emails and phishing attacks are growing rapidly. Traditional systems only classify spam but do not provide:

• —Risk level assessment
• —Confidence scoring
• —Explanation

This project solves that by building a complete AI environment.

💡 Solution

This system:

• —Detects spam emails
• —Classifies risk (Low / Medium / High)
• —Gives confidence score
• —Provides explanation
• —Simulates agent learning with multi-step tasks

⚙️ Features

• —✅ Spam Detection (ML based)
• —✅ Risk Level Classification
• —✅ Confidence Scoring
• —✅ Explanation Generation
• —✅ Multi-Step OpenEnv Environment
• —✅ FastAPI Backend
• —✅ Docker Deployment
• —✅ Live Hosted on Hugging Face

🏗️ System Architecture

• —Model → TF-IDF + Naive Bayes
• —Backend → FastAPI
• —Environment → Custom OpenEnv
• —Deployment → Docker + Hugging Face

🤖 OpenEnv Tasks

🟢 Easy

• —Detect spam or not

🟡 Medium

• —Classification + confidence

🔴 Hard

• —Risk level + explanation

👉 Reward is calculated based on:

• —Accuracy
• —Confidence
• —Risk detection

🔌 API Endpoints

🧪 Example Input

{
  "text": "Win a free lottery now!!!"
}

📊 Example Output
JSON
{
  "intent": "spam",
  "confidence": 0.92,
  "risk_level": "high",
  "explanation": "Suspicious or harmful content detected",
  "score": 1.0
}

🐳 Deployment
This project is fully containerized using Docker and deployed on Hugging Face Spaces.

📦 Tech Stack
Python
FastAPI
Scikit-learn
Docker
Hugging Face Spaces

👨‍💻 Author
Spandan Satapute
CSE Student | AI Developer