This application uses gemini as llm
A better version of RAG is implemented: https://github.com/deb-cod/multi-file-rag
Here, vecotr db (Qdrant) doker, local llm: Laama 3 8B model is used, llm accessed using fast-api, watcher (watchdog) for. file. change