We are looking for an experienced freelancer to set up a fully functional Retrieval-Augmented Generation (RAG) application using Open WebUI as the chat interface. The project involves configuring and integrating multiple components into a cohesive, production-ready pipeline for document-based question answering. Tech Stack & Architecture The application stack is as follows: • Chat UI: Open WebUI • LLM Runtime: Ollama (Llama 3.1 70B or Qwen 2.5 72B) • Embedding Model: BGE-M3 (external service) • Vector Database: Qdrant • Ingestion Pipeline: LlamaIndex • Re-ranking: BGE-Reranker or Cohere Rerank • RAG Evaluation: RAGAs • Fine-tuning Loop: DPO on collected reference data (periodic)