🛡️ Aura-Sentinel

---

Enterprise AI-Powered Customer Retention Platform

Combining XGBoost + Deep Q-Network Reinforcement Learning + Real-time Analytics

Features • Quick Start • Architecture • AI Models

---

🎯 What is Aura-Sentinel?

Aura-Sentinel is an enterprise-grade AI platform for customer churn prediction and retention optimization. It helps businesses:

• 📉 Predict which customers are likely to churn
• 🎯 Decide the optimal retention action for each customer
• 💰 Maximize revenue saved while minimizing intervention costs
• 📊 Visualize real-time analysis in a modern dashboard

---

✨ Features

🎯 Churn Prediction XGBoost model with 94% accuracy predicting customer churn probability 🤖 RL Action Selection Deep Q-Network agent optimizes retention actions (Email, SMS, Discounts, Personal Call) 🔮 Oracle Mode What-if scenario analysis - adjust cost modifiers to see how AI decisions change

📊 Live Matrix Feed Real-time customer processing with animated visualization 🧪 Training Lab Upload custom datasets and train new models with one click 📋 Reports & Export Filter by risk level, export to PDF and CSV

---

🏗️ Architecture

┌─────────────────────────────────────────────────────────────────────┐
│                    AURA-SENTINEL DESKTOP APP                        │
├─────────────────────────────────────────────────────────────────────┤
│                                                                     │
│    ┌──────────────┐     ┌──────────────┐     ┌──────────────┐       │
│    │   FRONTEND   │     │   BACKEND    │     │   AI BRAIN   │       │
│    │              │     │              │     │              │       │
│    │  React 18    │◄───►│   Go 1.21    │◄───►│  Python 3.10 │       │
│    │  TypeScript  │     │   Wails 2.11 │     │  Flask API   │       │
│    │  Recharts    │     │   Bindings   │     │  PyTorch     │       │
│    │  Lucide      │     │              │     │  XGBoost     │       │
│    └──────────────┘     └──────────────┘     └──────────────┘       │
│                                                                     │
└─────────────────────────────────────────────────────────────────────┘

---

🚀 Quick Start

Prerequisites

Tool	Version	Install Command
Go	1.21+	Download
Node.js	18+	Download
Python	3.10+	Download
Wails CLI	2.11	go install github.com/wailsapp/wails/v2/cmd/wails@latest

1️⃣ Start Python Brain API

cd apps/brain-rl
python -m venv venv
.\venv\Scripts\activate    # Windows
pip install -r requirements.txt
python api.py

2️⃣ Run Wails Desktop App

cd apps
wails dev

3️⃣ Build for Production

cd apps
wails build

Output: build/bin/Aura-Sentinel.exe

---

📁 Project Structure

aura-sentinel/
├── 📂 apps/
│   ├── 📄 main.go           # Wails entry point
│   ├── 📄 app.go            # Engine bindings & API
│   ├── 📂 frontend/         # React TypeScript UI
│   │   ├── src/App.tsx      # Main dashboard component
│   │   └── src/App.css      # Premium dark theme
│   ├── 📂 brain-rl/         # Python AI models
│   │   ├── api.py           # Flask REST API
│   │   ├── *.pth            # PyTorch DQN weights
│   │   └── *.pkl            # XGBoost model
│   └── 📂 engine-go/        # Standalone batch processor
├── 📂 data/
│   └── dataset.xls          # Telco customer data
└── 📄 README.md

---

🧠 AI Models

XGBoost Churn Predictor

Metric	Value
Accuracy	94%
Features	22 customer attributes
Output	Churn probability (0.0 - 1.0)

Deep Q-Network (DQN) Agent

Component	Description
State	9 features (churn prob, tenure, charges, contract, etc)
Actions	6 retention actions with varying costs
Reward	Customer Lifetime Value saved - action cost
Network	4-layer MLP (128→128→64→6)

Available Actions

ID	Action	Cost
0	No Action	0%
1	Send Email	1%
2	Send SMS	2%
3	Offer 10% Discount	10%
4	Offer 20% Discount	20%
5	Personal Call + 30% Discount	35%

---

🔮 Oracle Mode

Adjust the cost modifier to simulate different business scenarios:

Modifier	Effect
0.5x	Discounts are cheaper → AI prefers discounts
1.0x	Normal business pricing
3.0x	Discounts are costly → AI prefers Email/SMS

This demonstrates how the RL agent adapts its strategy based on business constraints.

---

�️ Dashboard Pages

Page	Description
Dashboard	Live matrix feed, Oracle control, charts
Analytics	Retention trends, AI performance metrics
Training Lab	Upload datasets, train new models
Reports	Filter & export customer data

---

🛠️ Tech Stack

• Frontend: React 18, TypeScript, Vite, Recharts, Lucide Icons
• Desktop: Wails 2.11 (Go + WebView2)
• AI Backend: Python, Flask, PyTorch, XGBoost, NumPy
• Styling: Custom CSS with glassmorphism, dark theme

---

📝 License

MIT License - Free for personal and commercial use.

---

Built with ❤️ using Go, Python, and React

A modern AI-powered desktop application for enterprise customer retention