AI API

This project uses FastAPI to serve AI models as APIs.

Project Structure

To start a new project you just need to add your custom scripts to the repo. Please respect the naming conventions and format, in order for your project to run smoothly. Here's a simplified overview of the structure sentiment:

Dockerfiles/
    sentiment

workers/
    sentiment/
        worker.py
        requirements.txt

config_models/
    sentiment.py

handlers/
    sentiment.py

Getting started with a new project

Please refer to the project sentiment to have a reference template to work with.

• add your custom Dockerfile in Dockerfiles/
• add your custom worker in workers/, by creating a new directory
• add your model config in config_models/, by creating a new .py file
• add your custom handler in handlers/

Setup the API

Build and run the api

# build and enter the container
export AI_API_ENDPOINT_NAME=sentiment
sh build_and_run.sh $AI_API_ENDPOINT_NAME

# start the server
sh ./server/uvicorn.sh

# call it
curl http://127.0.0.1:8080

curl -X POST http://127.0.0.1:8080/sentiment \
    -H "Content-Type: application/json" \
    -d '{"input": "I am very happy today!"}' 

API Documentation

FastAPI automagically generates a documentation for your API

http://127.0.0.1:8080/docs

Add gunicorn

Refer to this repo https://github.com/loretoparisi/dserve to add support for gunicorn.