Skip to main content
🧠

ragviz

Graph RAG visualization platform

An interactive platform for visualizing Graph Retrieval-Augmented Generation (RAG) pipelines. Upload documents, build knowledge graphs, and chat with local LLMs while tuning parameters in real-time.

The Problem

RAG pipelines are opaque — developers can't see how documents are chunked, embedded, and retrieved. Debugging retrieval quality requires visibility into the knowledge graph structure.

The Approach

ragviz provides a visual interface to the entire RAG pipeline: document ingestion, graph construction with Neo4j, vector search with HNSW indexing, and interactive querying against local LLMs via Ollama and llama.cpp.

Results & Impact

  • Interactive visualization of 10K+ node knowledge graphs
  • Real-time parameter tuning for RAG pipeline optimization
  • Full local LLM support via Ollama and llama.cpp