Rosmarium COS: AI-Native Content Orchestration System

Rosmarium COS: AI-Native Content Orchestration System

Rosmarium COS: The AI-Native Content Orchestration System

Rosmarium COS is an open-source, AI-native headless content orchestration system built with a pragmatic split-stack architecture. It is designed to act as the central nervous system for modern digital experiences, integrating seamlessly with both frontend applications and asynchronous background AI workers.

Project Overview

Traditional Content Management Systems segment content storage from the increasingly important AI processing pipelines. Rosmarium COS bridges this gap natively. It provides a robust, strongly-typed foundation for content editors and developers while integrating efficiently with Python-based AI workers for tasks like vector embeddings, Knowledge Graph extraction, NER (spaCy), and Graph Analytics (NetworkX) out-of-the-box.

Technologies Used

  • TypeScript & Fastify: The high-performance core API server, providing the primary interface for content delivery.
  • GraphQL (Pothos & Yoga): A fully automatic, code-first GraphQL layer offering deep querying, mutations, and WebSocket subscriptions for real-time updates.
  • Drizzle ORM & PostgreSQL: A type-safe persistence layer leveraging pgvector for advanced AI similarity searches.
  • BullMQ & Redis: A robust asynchronous messaging layer powering the webhook delivery system and internal event pipelines.
  • Python (FastAPI): A dedicated, non-public AI worker pipeline powered by Ollama for vector embeddings, spaCy for NLP, and NetworkX for graph analytics.

Key Features

  • 🧠 Pragmatic Split-Stack Architecture: Optimized at every layer, utilizing TypeScript for the web/API tiers and Python for specialized AI and vector workloads.
  • ⚑ Automated GraphQL API: A robust implementation of a Relay-compliant connection system, ensuring client scalability with completely N+1 safe DataLoaders built directly in.
  • πŸ•ΈοΈ Real-Time Webhooks & Subscriptions: Event-driven architecture utilizing a transactional outbox and background queue system to guarantee webhook deliveries, alongside live WebSocket subscriptions.
  • πŸ€– Built-In AI Pipelines: Direct integration through queues for AI enrichment workloads, extracting rich Knowledge Graphs and semantic indexes automatically.