Productizing ML at Scale at Twitter with Yi Zhuang

800 800 The TWIML AI Podcast (formerly This Week in Machine Learning & Artificial Intelligence)

Today we continue our AI Platforms series joined by Yi Zhuang, Senior Staff Engineer at Twitter & Tech Lead for Machine Learning Core Environment at Twitter Cortex.

In our conversation, Yi guides us through the machine learning landscape at Twitter, starting with the history of the Cortex team, who are responsible for building and maintaining their ML platform. We dig into their recently released Deepbird v2 framework, which is used for model training and evaluation solutions, and it’s integration with Tensorflow 2.0. We also discuss their efforts in automating hyperparameter searches, pipeline management, and their newly assembled “Meta” team, that is tasked with exploring the bias, fairness, and accountability of their machine learning models.

Join us at TWIMLcon: AI Platforms

Today we’re super excited to announce the launch of our inaugural conference: TWIMLcon: AI Platforms! TWIMLcon: AI Platforms will focus on the platforms, tools, technologies, and practices necessary to scale the delivery of machine learning and AI in the enterprise.

You already know TWIML for bringing you dynamic, practical conversations via the podcast, and we’re creating our TWIMLcon events to build on that tradition. The event will feature two full days of community-oriented discussions, live podcast interviews, and practical presentations by great presenters sharing concrete examples from their own experiences.

By creating a space where data science, machine learning, platform engineering, and MLOps practitioners and leaders can share, learn and connect, the event aspires to help seed the development of an informed, sustainable community of technologists that is well equipped to meet the current and future needs of their organizations.

Some of the topics we plan to cover include:

  • Overcoming the barriers to getting machine learning and deep learning models into production
  • How to apply MLOps and DevOps to your machine learning workflow
  • Experiences and lessons learned in delivering platform and infrastructure support for data management, experiment management, and model deployment
  • The latest approaches, platforms, and tools for accelerating and scaling the delivery of ML and DL in the enterprise
  • Platform deployment stories from leading companies like Google, Facebook, Airbnb, and traditional enterprises like Comcast and Shell
  • Organizational and cultural best practices for success

The two-day event will be held on October 1st and 2nd at the Mission Bay conference center in San Francisco, and I’d really love to meet you there! Early bird registration is open today at and we are offering the first 10 listeners who register the amazing opportunity to get their ticket for 75% off using the discount code TWIMLFIRST!

Thanks to our Sponsor!

I’m really grateful to our friends over at SigOpt who stepped up to support this project in a big way. In addition to supporting our AI Platforms podcast series and next ebook, they’ve made a huge commitment to this community by signing on as the first Founding Sponsor for the event. SigOpt’s software is used by enterprise teams to standardize and scale machine learning experimentation and optimization across any combination of modeling frameworks, libraries, computing infrastructure and environment. Teams like Two Sigma, who we’ll hear from later in this podcast series, rely on SigOpt’s software to realize better modeling results much faster than previously possible. Of course, to fully grasp its potential it is best to try it yourself. This is why SigOpt is offering you an exclusive opportunity to try their product on some of your toughest modeling problems for free. To learn about and take advantage of this offer, visit!

About Yi

Mentioned in the Interview

“More On That Later” by Lee Rosevere licensed under CC By 4.0

Leave a Reply

Your email address will not be published.