TWIMLcon Day 3:
How to Get Models from Development to Production Consistently and Predictably

Day 3 of TWIMLcon 2012: AI Platforms was all about how to get models from development to production reliably and consistently by using modern MLOps tools and platforms.

(The conference started Tuesday and runs through Friday, January 29th. It’s not too late to join in! Use discount code GREATCONTENT for 25% off registration.)

David Hershey, Solutions Architect at Tecton led a three hour workshop on how to deploy a fraud detection model leveraging their Feature Store product. Why even use a Feature Store? In David’s words:

“Feature stores enable faster development cycles, lower time-to-production, lower operational costs, and easier adoption of ML across your teams.”

He started off with raw data, did some feature engineering, created a training dataset, trained the model, and created end-points for both the model and features for the model to use during inference. Finally he demonstrated integrating it all into production.

We had some fun with the networking today by playing a variety of high-speed games. We can’t tell you which games, in case you try to prepare for the next TWIMLcon!

Finally Kristopher Overholt, a Solution Engineer from Algorithmia, walked through the high-level issues facing any organization struggling with getting ML into production and finding the ROI of the project. Specifically, he called out some pretty common challenges:

“Getting stuck in the lab, disconnected teams, technology mismatch, lack of stakeholder buy-in, hidden technical debt, all leading to an inefficient machine learning lifecycle.”

We’d like to thank David and Kristopher for sharing their time and expertise with everybody both in their workshop presentations and in the Q&A.

The conference continues tomorrow, Day 4 of 8, with the Executive Summit. We will be hosting a line-up of great speakers from BP, Yelp, Booz Allen, Adobe, Accenture, Walmart, Ocelot Consulting, Carbon Health, Orangetheory Fitness, Prosus Group, and Cruise. Topics for the day will include:

  • Building the Business Case for ML Platforms;
  • Why AI projects Fail and How to Ensure Success;
  • Building Teams and Culture to Support ML Innovation.

If this sounds interesting, it’s not too late to register! There are still five more days of sessions, the Executive Summit above. Pro Plus and Executive passes provide ongoing access to the conference recordings so that you can catch up after the event. You can check out the agenda here and the speakers here.

Related Posts

Scroll to Top