TagsArticle Source
Deploy and Scale Models with BentoML
Abstract
This video is a recording from the ZenML Meet The Community session on 24th November 2022. The community session happens every Wednesday at 8:30 AM PT / 5:30 PM CET / 9:00 PM IST.
The goal is to engage the community and chat about ZenML’s latest features and showcase interesting demos or use cases. Sometimes we just take questions and have fun. Join us if you are curious about ZenML, or just want to talk shop about MLOps.
Deploying machine learning models is a tedious and time-consuming process. There are many moving parts involving multiple stakeholders in the journey to production.
The roles may vary from one team to another, but in a reasonable-scale ML project, you’d have - ‣ Data Scientist - Experimenting and creating accurate models. ‣ ML Engineer - Converting finalized experiments into an automated and repeatable process. ‣ DevOps Engineer - Deploying, serving, and monitoring models at scale. ‣ Product Manager - Delivering the product with a great user experience while balancing operational costs and resources.
Each stakeholder has a focus and also an important stake in the success of an ML project.
Today there’s a lot of friction in the process. This is why most ML models don’t make it into production.
ZenML wants to make this process easier and faster. Our new integration with BentoML makes the transition from experimentation to production seamless.
ZenML lets you connect your favorite tools in one place. BentoML lets you deploy and serve your models quickly.
⚡ In this video you’ll learn about-
‣ What is a Pipeline and a Step - How does it work in ZenML?
‣ What is a Model Deployer and the BentoML framework?
‣ How to run a YOLOv5 inference pipeline with BentoML?
‣ Deploying the YOLOv5 Bento on AWS Sagemaker.