MemLink - Stateful AI Integration Middleware

MemLink is a production-ready middleware that provides:

• Persistent session memory per user/device.
• Vector embeddings for semantic memory retrieval.
• Multi-session and overlay support for multiple accounts.
• Context-aware prompt generation for AI models.
• REST API interface via FastAPI.
• Containerized deployment via Docker/Docker Compose.

Features

1. Multi-session handling: Each user/device can have independent or shared session memory.
2. Persistent vector memory: Stores embeddings in FAISS for fast similarity search.
3. Overlay support: Multiple accounts can share a base session and apply personal overlays.
4. API endpoints:
   • GET / - Health check.
   • POST /query - Query session memory and get prompt.
5. Docker support: Ready for containerized deployment.
6. Integration ready: Can integrate with other projects (e.g., Flyometer).

Quick Start

1. Build Docker image:

docker-compose build

2. Start container:

docker-compose up

3. Query API:

POST http://localhost:8000/query
{
    "session_id": "user123",
    "user_input": "Explain stateful AI integration",
    "top_k": 3
}

Repo Structure

MemLink/
├── src/                  # Core middleware modules
├── api/                  # FastAPI application
├── data/                 # Session data storage
├── examples/             # Example scripts
├── Dockerfile            # Docker configuration
├── docker-compose.yml    # Compose file for container orchestration
├── requirements.txt      # Python dependencies
├── pyproject.toml        # Project metadata
└── README.md             # This file

License

MIT License