Data Council Blog

Data Council Blog

Open Source Highlight: Streamlit

Streamlit officially launched out of beta on October 1st, 2019 with the promise to "turn Python scripts into beautiful ML tools." On the same day, Google's AI-focused venture fund Gradient Ventures announced its investment into the startup, which has since then attracted a considerable amount of attention despite its young age.

Streamlit - the fastest way to build data apps

Source: Streamlit

In practical terms, Streamlit is an open source app framework that aims to make building ML tools as easy as writing Python scripts. While it may sound trivial, this actually solves a big pain point for teams doing machine learning. They have a constant need for tools to present and explore their results; but the process to come up with such tools was far from simple.

CEO Adrien Treuille was well-placed to identify and solve this issue: he had witnessed it first-hand at Carnegie Mellon, Berkeley, Google X, and autonomous driving startup Zoox (yes, his resume is impressive). Before Streamlit, the problem he saw would typically go two ways: either ML engineers and data scientists would get distracted by having to create and maintain tools themselves; or another team would get involved, once, and the tools wouldn't keep up with changes. With Streamlit, the workflow becomes much more manageable:

Streamlit - Open source framework for building ML apps

Source: Streamlit

This explains why even in its beta days, Streamlit's community already included engineers from Uber, Twitter, Stitch Fix, and Dropbox; and why it has been so quickly adopted. Only a few months after its launch, it has already facilitated "the emergence of an intricate archipelago of Streamlit apps, from simple, fun demos of open-source projects, to community-spanning infrastructure and complex internal tooling at major companies used in production by hundreds of employees," Treuille noted.

If you are keen to hear more about Streamlit, check out episode #238 of Python-related podcast Podcast.__init__, in which host Tobias Macey interviewed Adrien Treuille.

Data Science, Data Visualization, Machine Learning, Open Source