MLflow Integration from Azure ML and Algorithmia

Join us for virtual tech talks at Data + AI Meetup about MLflow Integration from Azure ML and Algorithma sponsored by the Databricks MLflow Team. It will be simultaneously broadcasted live on YouTube and LinkedIn.

Intro Slides, MLflow Community Info: https://databricks.slack.com/files/U0...

Agenda:
9:00 - 9:05 AM: Introduction & Announcements
9:05 - 9:35 AM: MLflow + Azure ML Integration
9:40 - 10:10 AM: Deploy Models to Algorithmia using MLflow

Quick links:
MLflow: https://mlflow.org/
Azure ML: https://docs.microsoft.com/en-us/azur...
Algorithma: https://algorithmia.com/

Talk One

Title: MLflow + AzureML
Presenter: Eduardo de Leon
Abstract: AzureML has a long history of providing solutions for the ML Lifecycle and now offers a vendor-agnostic approach for Data Scientists and ML Engineers to integrate with our enterprise-grade services via MLflow. Operationalizing MLflow in production offers some unique challenges, and we’ll talk about mlflow-skinny - a slimmed-down, system-integrator friendly MLflow client - today as one part of that puzzle.

Links for the integration:
Examples: https://github.com/Azure/MachineLearn...
Azure/MachineLearningNotebooks:
https://github.com/Azure/azureml-exam...
MLflow Skinny : https://pypi.org/project/mlflow-skinny/
PRs: https://github.com/mlflow/mlflow/pull... https://github.com/mlflow/mlflow/pull...


Bio: Eddie is a Software Engineer on the AzureML team and has learned about building composable and maintainable SDKs the hard way, having developed the core of the initial AzureML Python SDK. Bringing those learnings to the MLflow community and AzureML users and applying them to differentially private scenarios is what makes Eddie the happiest!

Talk Two

Title: Deploy models to Algorithmia using MLflow
Presenter: Daniel Rodriguez
Abstract: We will take a look at the Algorithmia (algorithmia.com) platform. An MLOps platform allows data scientists to put their models in production by wrapping any machine learning inference code into a REST API and helps users scale and manage these deployments in an easy and convenient way without having to think of containers, kubernetes, and governance. We will also look at how MLflow helps users abstract different parts of the Machine Learning lifecycle, including deployment, and how we built an MLflow plugin to help users deploy any MLflow projects and experiments to Algorithmia. Databricks is proud to announce that Gartner has named us a Leader in both the 2021 Magic Quadrant for Cloud Database Management Systems and the 2021 Magic Quadrant for Data Science and Machine Learning Platforms. Download the reports here. https://databricks.com/databricks-nam...