Strange Loop

Scalable Machine Learning Pipelines with MLflow

Developing Machine Learning Products that can scale comes with numerous barriers. From gathering data, training a model, and deploying the model, there is a complex series of steps needed to be done at each step to provide meaningful output. With MLflow we can encapsulate the machine learning pipeline akin to a Docker image to reduce the manual processes needed to deploy a model.

In this workshop we will use MLflow to build a machine learning pipeline. We will explore how MLflow allows us to create scalable, reproducible, and trackable machine learning pipelines through its 3 main modules.

MLflow Tracking- Tracks experiments to record and compare parameters and results MLflow Projects - Packages ML code in a reusable, reproducible form to share or transfer to production MLflow Models - Manages and deploys models from a variety of ML libraries to a variety of model serving and inference platforms.

This Is an intermediate workshop that will require knowledge of using python and navigating a Linux terminal. No machine learning experience is necessary.

Banjo Obayomi

Banjo Obayomi

Two Six Labs

Banjo is a Senior Research Engineer at Two Six Labs, where he develops platform solutions for productizing various researched based projects. Banjo received his B.S in Computer Science from University of Maryland College Park in 2011, and his M.S in Computer Science from Loyola University Maryland in 2015. Banjo also is an AWS Certified Solutions Architect - Associate and AWS Certified Big Data Engineer.