Centralized-LLM-Client

Unified Python client for interacting with multiple LLM providers (OpenAI, Gemini, OpenRouter) with automatic fallback, secure key management, file handling, and MongoDB logging.

Features

• Automatic provider fallback (OpenAI, Gemini, OpenRouter)
• Secure API key management (encrypted, MongoDB Atlas)
• File handling: supports local, S3, and FastAPI UploadFile
• Detailed logging of all LLM calls to MongoDB Atlas
• S3 integration for file storage/retrieval

Installation

Python 3.9+ required.

pip install Centralized-LLM-Client
# Or, for local development:
pip install -r src/requirements.txt

Quickstart

from src.CentralizedClient import CentralizedLLMClient
client = CentralizedLLMClient(username="alice")
response = await client.completion("Summarize this document.")

Architecture

• src/CentralizedClient/CentralizedLLMClient.py: Main orchestration, provider fallback logic
• Providers: gemini_provider.py, openai_provider.py, openrouter_provider.py (wrap LLM APIs)
• Key Management: key_manager.py (encrypted keys, user/system fallback)
• File Handling: file_processor.py (temp files, S3/local/FastAPI)
• Logging: llm_logger.py (MongoDB Atlas logging)
• External Integrations: S3 (utils/s3_file_manager.py), MongoDB Atlas (utils/atlas_client.py)

Environment Setup

• Set environment variables for S3, MongoDB Atlas, and LLM provider keys (see .env usage in utils)
• Place key.bin (encryption key) in project root for key decryption

Testing

• Add tests in tests/ and run with pytest:

   pytest tests/
  

Contributing

Pull requests are welcome! For major changes, please open an issue first to discuss what you would like to change.

License

MIT License