Fast, documented Machine Learning APIs with FastAPI

Video description

Use FastAPI to expose an HTTP API for fast live predictions using an ONNX Machine Learning Model. FastAPI is a Python web framework that provides easy development of documented HTTP APIs by offering self-documented endpoints with Swagger - a tool to describe, document, and use RESTful web services.
Learn how to quickly put together an API which validates requests, and self-documents its endpoints using OpenAPI via Swagger. Quickly produce a robust interface for others to consume your Machine Learning model by following core best-practices of MLOps.
Parts of this video cover the basics of packaging Machine Learning models, as covered in the Practical MLOps book.
Topics include:
* Create a Python project to serve live predictions using FastAPI
* Use a Dockerfile to package the model and the API using Docker containerization
* With minimal Python code, expose an ONNX model to perform sentiment analysis over an HTTP endpoint
* Dynamically interact with the API using the self-documented endpoint in the container.

Useful links:
* Demo Github Repository with sample code
* Practical MLOps book
* FastAPI Intro tutorial
* RoBERTa ONNX Model for sentiment analysis

Table of contents

  1. Lesson 1
    1. "Onnx With Fastapi"

Product information

  • Title: Fast, documented Machine Learning APIs with FastAPI
  • Author(s): Alfredo Deza, Noah Gift
  • Release date: July 2021
  • Publisher(s): Pragmatic AI Solutions