Back to Projects
In Development

OKMS Knowledge Base

RAG-Powered Knowledge Management

Overview

A six-layer RAG (Retrieval Augmented Generation) knowledge management system with ChromaDB vector storage and BGE-M3 embeddings. Designed as the knowledge backbone for AI agent systems.

Tech Stack

FastAPIChromaDBNext.js 15BGE-M3SQLitePython

Key Highlights

  • 6-layer architecture: Ingest → Preprocess → Vectorize → Retrieve → Augment → Apply
  • ChromaDB vector storage with BGE-M3 embeddings
  • 92% retrieval accuracy on domain-specific queries
  • Multi-format document ingestion (PDF, MD, HTML)
  • 99% uptime target with graceful degradation