🔗
RAG Pipeline
Upload PDFs and automatically extract, chunk, embed, and store content using a full Retrieval-Augmented Generation pipeline powered by Qwen2.5 and pgvector.
⚡
CPU-Only Inference
Runs entirely on commodity hardware — no GPU required. Ollama serves Qwen2.5-1.5B for embeddings and inference on a 64 GB Hetzner server.
🌍
7 Languages
Full UI localization in Czech, English, German, French, Spanish, Russian, and Chinese. JWT authentication with complete audit logging.