I built NovaNews, an AI‑driven news aggregator that runs as three independent microservices – a Angular 21 frontend, a Node.js/Express/MongoDB backend with JWT authentication, and a Python/Flask AI service using LangChain, Groq’s Llama 3.3 70B model, ChromaDB, HuggingFace embeddings and newspaper3k.
The frontend is implemented with Angular 21.
The backend uses Node.js, Express, MongoDB and JWT for authentication.
The AI service runs on Python/Flask, leveraging LangChain, the Groq Llama 3.3 70B model, ChromaDB for vector storage, HuggingFace embeddings and newspaper3k for scraping.
NovaNews provides RAG‑based Q&A over collected articles, batch summarization, and smart question suggestions.
The system is described as "complete", meaning all intended features are implemented and operational. For a related project that also uses a microservice‑like architecture, see my previous news aggregator.