Create a Vector Database and learn about LLMs

This repository is designed for a workshop about learning vector database, LLMs, RAGs and so on.

To interact with the LLM portion, you will need access to an OpenAI API key or Google Gemini service account. In the /config/config.env file, add your information as follows:

• For OpenAI

  PROJECT_ID="Your projectID if needed"
  LOCATION=
  OPENAI_API_KEY="your OpenAI API key"
  MODEL_ID="the model to use, e.g. gpt-4o"
  PATH_TO_AUTH=
  

• For Google Gemini

  PROJECT_ID="Your projectID"
  LOCATION="Your project location, e.g. us-central1"
  OPENAI_API_KEY=
  MODEL_ID="the model to use, e.g. google/gemini-1.5-flash-002"
  PATH_TO_AUTH="path to your auth.json file"
  

There are two ways to interact with this repo:

With containers

1. Build and deploy the containers:

   DOCKER_BUILDKIT=0 docker compose up -d
   

2. Access the frontend at http://localhost:4173.
3. Use an IDE, like VS Code, to connect to the running container named server.
4. Navigate to the /server/ directory to interact with the backend code.

Without containers

Note

Python 3.10, Node.js 22, and Yarn 1.22 are required to run this demo

Be sure to export the necessary environment variables from above to interact with the LLM.

1. From the /website/ directory run:

   yarn && yarn dev
   

   Access the frontend at http://localhost:4173.
2. From the /server/ directory run:

   uvicorn main:app --host 0.0.0.0 --port 8080 --reload