LLM_Data_Analyst

Setup

• Navigate to the backend folder and start the FastAPI server using uvicorn app:app --reload
• Navigate to the frontend folder and call npm install followed by npm run dev
• The backend folder will require a sandbox.py file that includes the variable API_KEY, which links to Google Gemini.

Overview

This is a prototype LLM-based React app that allows users to upload files and have Google Gemini interact with them. The app was built with a number of React libraries, Python (FastAPI), and Google Gemini's Python API.

Technical Specifics

• The frontend and backend interact through Python's FastAPI module.
• The system instructions can be read in the app.py file.
• The app can only interact with CSV files. It outputs its response in markdown, which is then formatted by the frontend.
• The app uses UUIDs to uniquely identify each chat; therefore, you can leave a chat and return to it through the notebook button on the right side of the screen. These chats do not persist after a restart as they are currently only held in a dictionary in memory.