AI-Enhanced Group Chat (Web App)

Overview

This project is a real-time group chat application with AI Assistance.

Multiple users can join the same chatroom, send messages, and see each other’s responses in real time.
Any message ending with a ? automatically triggers a response from an AI assistant using Mistral’s models.
Each AI response can be liked or disliked by users, and the collected feedback is used in a preference fine-tuning loop with GRPO to continuously improve model behavior.

---

Features

• 🔐 User authentication (JWT-based)
• 👥 Shared group chatroom — all users see each other’s messages
• 💬 Real-time chat via WebSocket
• 🤖 AI bot replies (triggered by ?) powered by Mistral
• 👍👎 Feedback system — like/dislike buttons on each AI response
• 🔄 Feedback-to-fine-tuning loop using GRPO for preference adaptation
• 📦 Clear, modular structure with .env configuration
• 🛠 Easy to set up and test locally or in deployment

---

Prerequisites

• OS: Linux, macOS, or Windows
• MySQL 8+
• Python 3.10+ and pip

---

(Option 1) Setup Instructions (To run locally)

1. Database Setup (MySQL)

CREATE DATABASE groupchat CHARACTER SET utf8mb4 COLLATE utf8mb4_unicode_ci;
CREATE USER 'chatuser'@'localhost' IDENTIFIED BY 'chatpass';
GRANT ALL PRIVILEGES ON groupchat.* TO 'chatuser'@'localhost';
FLUSH PRIVILEGES;

---

2. Backend (FastAPI)

cd backend
python -m venv .venv
source .venv/bin/activate   # Windows: .venv\Scripts\activate
pip install -r requirements.txt

Create .env and paste the following:

DATABASE_URL=mysql+asyncmy://chatuser:chatpass@localhost:3306/groupchat
JWT_SECRET=<replace_with_long_random_string>
JWT_EXPIRE_MINUTES=43200
LLM_API_BASE=https://api.mistral.ai/v1
LLM_MODEL=mistral-medium
LLM_API_KEY=<your_mistral_api_key>
APP_HOST=0.0.0.0
APP_PORT=8000

Run the server:

uvicorn app:app --host 0.0.0.0 --port 8000

Access the app at: http://localhost:8000

(Option 2) 🚀 Run with Docker (Recommended)

This setup lets you run the FastAPI + MySQL + Frontend stack using Docker Compose — no manual environment setup required.

---

1️⃣ Build and Run Containers

From the project root:

docker compose up --build

This command will:

• Build the FastAPI app image from your Dockerfile

• Launch MySQL 8.0 and the FastAPI backend

• Automatically initialize the database from sql/schema.sql

Once complete, you’ll see:

Uvicorn running on http://0.0.0.0:8000

---

3. Frontend

The frontend is HTML/CSS/JS served by FastAPI:

• REST API → login, signup, and posting messages
• WebSocket → broadcasting chat messages in real time
• Feedback UI → like/dislike buttons on AI messages

---

Usage

1. Start the FastAPI server (uvicorn ...).
2. Open http://localhost:8000 in a browser.
3. Sign up for an account and log in.
4. Share the same URL with other users — each can sign up and join.
5. All logged-in users connect to the same group chatroom.
6. Messages are broadcast in real time to all participants.
7. Any message ending with ? will trigger an AI response from Mistral.
8. Users can like or dislike AI responses.
9. Feedback is stored in the DB and used in a GRPO fine-tuning pipeline to improve model responses over time.

---

Project Structure

backend/
  ├── app/                # FastAPI app
  ├── requirements.txt    # Python dependencies
  └── .env.example        # Example environment configuration
frontend/
  ├── static/             # HTML, CSS, JS files
sql/
  └── schema.sql          # DB schema setup
README.md

---