Machine learning model serving with streamlit and FastAPI


#1

Davide Fiocco

Abstract:
Python provides sophisticated frameworks to develop machine learning models that can perform a variety of useful tasks. To put those models to good use, one frequently needs to encapsulate them in web applications, so to make them available to human users and other applications. I will show how a dockerized web application serving a ML model can be coded in pure Python, building its frontend with the streamlit library and the backend with the FastAPI library (no Javascript, no HTML/CSS needed!).

Bio:
Currently I am working as senior data scientist at Frontiers (https://www.frontiersin.org/) in Madrid. There, I develop and help bring to production machine learning algorithms consuming scientific text and images, with the ultimate goal of facilitating the publication of high quality scientific content.
I contribute to Python open source projects as @davidefiocco on GitHub and I started blogging Python things on https://davidefiocco.github.io/. I also overuse my Twitter handle @monodavide.