Unified LLM inference proxy with multi-provider routing, fallbacks, and caching - View it on GitHub