TinyLLM

A lightweight Python framework for building, testing, and deploying production-grade LLM applications.

TinyLLM moves AI logic from experimental scripts to production-grade services. It organizes LLM workflows into modular units (chains and tools) to ensure every prompt and response is traceable. The system manages the heavy lifting for RAG pipelines and multi-agent systems while integrating with core providers like OpenAI and Anthropic. Use its built-in evaluation hooks to maintain high output quality: a critical requirement for shipping reliable code without the bloat of larger alternatives.

https://github.com/pieroit/tinyllm