A Full Information to Serving a Machine Studying Mannequin with FastAPI
Welcome to this third article in my MLOps collection. In the primary article, we explored Docker and the way it simplifies software packaging. In the second article, we managed machine studying fashions utilizing MLflow, Azure, and Docker. Now, on this third half, we’ll deliver all the things collectively by constructing a FastAPI software that serves our beforehand saved mannequin on Azure. This enables us to create a prediction service that may be accessed globally!
An API is sort of a bridge. Everytime you work together with a library in Python, you’re utilizing its API. It’s the general public a part of an software that you could work together with, whereas all the things behind it’s hidden.
APIs are generally used to speak with internet functions, and so they present a set of URLs that return information (You ship a request with some parameters and get a response again). Most frequently, the information comes again in codecs like JSON or XML, that are simple to parse. That is completely different from web sites that return HTML, which incorporates data for rendering pages. With APIs, you get simply the uncooked information.
Some APIs are public, whereas others are non-public. When constructing an API, you resolve what information to share, …
