Important

THIS REPOSITORY IS NOT UNDER ACTIVE DEVELOPMENT!

LLM Chat Wrapper

This project is an AI model wrapper developed for the Bilkent University CTIS221 - Object-Oriented Programming course. It provides a modular and extensible framework for integrating and managing various AI models, including support for chat, embeddings, and other capabilities.

Features

• Modular Architecture: Built with Spring Modulith for modularity and scalability.
• Database Integration: Uses PostgreSQL for persistent storage.
• Caching: Redis is used for caching frequently accessed data.
• Streaming Support: Real-time streaming capabilities using WebFlux and test streaming with Hono.
• AI Model Integration: Supports integration with various AI models and providers.

Prerequisites

• Java 17 or higher
• Gradle (for building the backend service)
• Docker (for running PostgreSQL, Redis, and other services)
• Bun (for the stream-hono service)

Setup Instructions

Backend (LLM Chat Service)

1. Clone the repository:

   git clone https://github.com/sezrr/ctis221-project.git
   cd ctis221-project/projects/llm-chat-service

2. Start the required services using Docker Compose:

   docker-compose -f compose.yaml up -d

3. Build and run the Spring Boot application:

   ./gradlew bootRun

4. Access the application at http://localhost:8080.

Stream Hono - OPTIONAL

1. Navigate to the stream-hono directory:

   cd ctis221-project/projects/stream-hono

2. Install dependencies:

   bun install

3. Start the development server:

   bun run dev

4. Open http://localhost:3000 in your browser.

Frontend (JavaFX GUI)

• Add instructions for running the JavaFX GUI.

API Documentation

The API documentation is available at:

• http://localhost:8080/swagger-ui.html for the backend services.

• MORE DOCUMENTATION

Project Structure

The project uses a monorepo style structure with the following projects:

• llm-chat-service: Spring Boot Modular Monolith Backend service for managing AI models, users, chats and their configurations.
  • aimodel: AI model management module.
  • chat: Chat module for interacting with AI models, and CRUD operations for chat history.
  • authentication: User authentication and authorization module.
  • instruction: Instruction module for managing instructions.
  • tiering: Tiering module for managing user tiers.
  • shared: Shared module for common utilities and configurations.
• stream-hono: Backend service for real-time streaming and interaction.
• modulo-ai: Core library for AI model integration and provider management.
• javafx-gui: JavaFX GUI for interacting with the backend services.

Technologies Used

• Spring Boot: Backend framework.
• PostgreSQL: Database.
• Redis: Caching.
• Hono: Real-time streaming.
• Bun: JavaScript runtime for the frontend.
• Docker: Containerization.

Team Members

• Arda Becanım
• Kerem Coşkun
• Sezer Tetik

Contributing

Contributions are welcome! Please follow these steps:

1. Fork the repository.
2. Create a new branch for your feature or bug fix.
3. Commit your changes and push them to your fork.
4. Submit a pull request.