Building an Open Source MLOps Stack with ZenML // Hamza Tahir // MLOps Workshop Meetup # 92

MLOps Community Workshop. Hamza Tahir CTO of ZenML gave a workshop about Building and Using an open source MLOps Stack with ZenML co-hosted by Ben Epstein. //Abstract Ever wanted to build an open source MLOps stack that lets you set up a pipeline with scheduled orchestration, drift detection, and automatic metadata tracking for reproducibility? In this talk, Hamza will create a real-world example of a reproducible ML pipeline using ZenML, and showcase how ZenML helps you deploy the pipeline using your favorite tools such as MLFlow, Kubeflow, Evidently, and more. // Bio Hamza Tahir is a software developer turned ML engineer. An indie hacker by heart, he loves ideating, implementing, and launching data-driven products. His previous projects include PicHance, Scrilys, BudgetML, and you-tldr. Based on his learnings from deploying ML in production for predictive maintenance use-cases in his previous startup, he co-created ZenML, an open-source MLOps framework to create reproducible ML pipelines. // Related links https://zenml.io ZenML NBA Pipeline Repository: https://github.com/zenml-io/zenfiles/ Blog: https://blog.zenml.io/zenhack-three-pointer-prediction/ ----------- ✌️Connect With Us ✌️------------- Join our Slack community: https://go.mlops.community/slack Follow us on Twitter: @mlopscommunity Sign up for the next meetup: https://go.mlops.community/register Catch all episodes, Feature Store, Machine Learning Monitoring and Blogs: https://mlops.community/ Connect with Demetrios on LinkedIn: https://www.linkedin.com/in/dpbrinkm/ Connect with Ben on LinkedIn: https://www.linkedin.com/in/ben-epstein/ Connect with Hamza on LinkedIn: https://www.linkedin.com/in/hamzatahirofficial/ Timestamps: [00:00] Introduction to Hamza Tahir [02:40] Full disclosure [03:15] Ben Epstein's background [05:06] Introduction to ZenML and Hamza Tahir's bg [10:45] Dependency issue [11:45] Initializing [12:17] Chapter 1 - Exploring NBA Data [14:34] Defining and creating a step [16:47] Defining and creating a pipeline [18:02] ZenML + Evidently [20:00] Running a pipeline [24:38] Post execution: Fetching pipelines and reviewing results [28:00] Caching mechanism [31:12] Chapter 2 - Training Pipeline [38:02] Pipeline in MLfow [41:47] ZenML Stacks [42:45] Goal of ZenML [45:06] Creating a stack [57:31] Chapter 3 - The Prediction Pipeline [1:08:05] Answering questions [1:08:53] Using integrations in building ZenML pipeline [1:13:00] ZenML Support [1:14:00] Hosting an ML application and incorporating ZenML [1:15:11] Version Control Management [1:18:06] Hamza's closing remark and shoutouts [1:19:32] Synchronous or asynchronous?