NVIDIA/nim-anywhere

Provides a complete RAG pipeline framework using NVIDIA NIM microservices (LLM, embedding, and reranker models) that scales from local development to production environments. Combines a Python-based chain server backend with a web frontend for knowledge base integration and chat inference, supporting flexible configuration via files or environment variables. Designed for AI Workbench but operates standalone, enabling enterprises to build private RAG systems that keep sensitive data local while leveraging optimized NVIDIA models.

209 stars. No commits in the last 6 months.