University of Minnesota
Software Engineering Center

You are here

ML Practice with ML Ops Principles

Date of Presentation: 
Thursday, January 16, 2020
Presented By: 
Machine learning (ML) has achieved considerable success in recent years and many companies have started using it in Production. Every company wants to get on this bandwagon to stay competitive and is challenged not only to choose the right technology but also to optimize the model by continuously deploying and observing. This talk will start from a strategic direction that we took at Virtuwell. We will cover how we leverage MLflow to manage the ML lifecycle, including experimentation, reproducibility, and deployment. The biggest challenge is how to get your data analysts, data scientists, software engineers and DevOps engineers to work in an agile fashion to pre-process data, create models, modify key data sets and deploy them to production and continuously optimize it. We will share the key lessons we learned on moving from DevOps to MLOps. Virtuwell is a 24/7 online clinic that helps customers get treatment plans for everyday illnesses and has increased its user base multi-fold year after year, for the past 9+ years.