Past Meetup

MLflow Meetup with Edmunds and Brandless @ Databricks HQ, SF

This Meetup is past

221 people went

Databricks

160 Spear, 13th Floor · San Francisco, CA

How to find us

The meetup will be on the 14th floor.

Location image of event venue

Details

Let’s kick off the New Year 2019 with our first Bay Area MLflow Meetup!

Join us for an evening of tech-talks about MLflow. Come and find out about new features, what’s coming in 2019, and hear from Brandless and Edmunds about how they use MLflow.

Due to building security requirements, you'll have to provide a government-issued ID to enter the premises.

Agenda:

6:00 - 6:30 pm: Social Hour with Food, Drinks, Beer & Wine
6:30 - 6:35 pm: Introduction & Announcements
6:35 - 7:10 pm: What’s Coming in MLflow v0.9 and v1.0 (Databricks)
7:10 - 7:35 pm: MLflow at Brandless
7:35 - 8:05 pm: MLflow at Edmunds
8:05 - 8:30 pm: Additional Networking & Q&A

Title: What’s Coming in MLflow v0.9 and v1.0

Presenters: Matei Zaharia and the MLflow Engineering Team
Abstract:
MLflow is aiming to stabilize its API in version 1.0 this spring and add a number of other new features. In this talk, we'll share some of the features we have in mind for the rest of the year. These include ongoing work such as a database store for the tracking server and Docker project packaging, as well as new improvements in multi-step workflows, model management and production monitoring.

Bio:
Matei Zaharia is an assistant professor of computer science at Stanford University and Chief Technologist at Databricks. He currently tech-leads the MLflow project at Databricks. Previously, Matei started the Apache Spark project during his Ph.D. at UC Berkeley in 2009 and also worked on other open source systems, such as Apache Mesos and Apache Hadoop.

Title: MLflow at Brandless

Presenters: Bing Liang and Adam Barnhard
Abstract:
Brandless is an e-commerce company with the intention of making better stuff accessible and affordable for more people. We use product information and purchase history to train a variety of recommendation engines to personalize our site for each customer. We use MLflow to manage the model deployment lifecycle: training the model, offline validating, deploying to production and tracking different model versions for live experimentation. We will discuss this process and share some key learnings that we've had along the way.

Bio:
Bing Liang currently works as a data scientist at Brandless, where she focuses on developing and implementing algorithms into Brandless production systems. Her specialties include recommendation and personalization systems, logistics optimization, and customer feedback analysis. She brings a background in engineering and information systems to the Brandless team.

Title: MLflow at Edmunds

Presenter: Ebraheem I. Fontaine
Abstract:
Edmunds receives millions of unlabeled vehicle photos daily from our dealer partners. We need to execute several image classification tasks to enhance these photos with valuable metadata. After building some initial proof-of-concept models, we migrated to the MLflow framework which has organized workflows and facilitated collaboration.

Bio:
Ebraheem I. Fontaine currently leads the Data Science Team at Edmunds and is focused on building predictive models that enhance our consumer and dealer products. Prior to Edmunds, he built models for payment card fraud detection and network intrusion detection at FICO. He earned a Ph.D. from Caltech in 2008 doing visual motion tracking for genetic model organisms.