Sightcall Q&A API

A RAG-based API to answer questions about SightCall using curated data.

Stack:

• FastAPI for the API, pytest for automated testing
• PGVector for the vector database allowing document retrieval
• Github Actions for CI/CD pipeline
• Google Cloud Run for containerized deployment
• Terraform for automated infrastructure provisioning

Installation

Prerequisites

• curl
• Docker Compose
• make
• uv

Setup

1. First, clone and install the project:

curl -sSL https://raw.githubusercontent.com/cmnemoi/sightcall_qa_api/main/clone-and-install | bash

2. Configure your OpenAI API key:

   • Open the .env file that was created during installation
   • Find the OPENAI_API_KEY variable
   • Replace sk- with your actual OpenAI API key

   You can get an API key from OpenAI's API keys page if you don't have one already.

Usage

You can interact with the API by running the following command:

curl -X POST "http://sightcall-qa.localhost/chat" \
     -H "Content-Type: application/json" \
     -d '{
           "question": "What is SightCall?"
         }'

Or by accessing the API OpenAPI / Swagger documentation at http://sightcall-qa.localhost/docs.

Indexation

To improve the RAG model's responses, you can index SightCall's website into the vector database by running:

make indexation

Development

• Lint code with make lint
• Run tests with make test
• Start the API locally with make watch

License

The source code of this repository is licensed under the AGPL-3.0-or-later License.