Strategy & Principles to Scale and Evolve MLOps @DoorDash

QCon London 2023

Session Machine Learning

Strategy & Principles to Scale and Evolve MLOps @DoorDash

Wednesday Mar 29 / 10:35AM BST, Mountbatten (6th Fl.)

Abstract

MLOps has become a major enabler to successfully operationalize ML applications and for ML practitioners to realize the power of ML to bring impact to business.  The journey to implementing MLOps will be unique to each company. One common thread to successfully adopt MLOps is the need for a strategy and a set of principles. 

At DoorDash, we’ve been applying MLOps for a couple of years to support a diverse set of ML use cases, such ETA predictions, the Dasher assignments, personalized recommendation of restaurants and menu items and more, and to perform large scale predictions at low latency.

This session will share our approach to MLOps, the strategy and principles that have helped us to scale and evolve our platform to support hundreds of models and billions of predictions per day, and deep dive on the technical aspects of scaling our feature store and prediction service.

Interview

I am currently leading the ML platform team at DoorDash. Our platform journey started a few years back and we are currently working on the second floor of our ML platform to support the various large and complex ML use cases that require distributed model training, model prediction flexibility and more

The use cases I mentioned are about recommendations, NLP, computer vision, and large language models.

At the previous QCon SF in 2022, one interesting thing I learned while attending the lightning talk given by Courtney Kissler, CTO @ Zuliy at the Women & Allies in Tech Breakfast was about using the criteria when making impactful decisions.

Topics

Machine Learning operations scalability
76% senior dev or higher
1:11 speaker ratio
60+ practitioners

QCon London 2023 is a three day conference for senior software engineers, architects and team leads. An international program committee of working engineers selects every session. Patterns and practices, not products and pitches.

Share

From the same track

Wednesday 29 March

10:35 Mountbatten (6th Fl.) Session Machine Learning Strategy & Principles to Scale and Evolve MLOps @DoorDash Hien Luu Sr. Engineering Manager @Zoox & Author of MLOps with Ray, Speaker and Conference Committee Chair 11:50 Mountbatten (6th Fl.) Session Digital Twins Cognitive Digital Twins: A New Era of Intelligent Automation Yannis Georgas Intelligent Industry Lead @Capgemini 13:40 Mountbatten (6th Fl.) Session AI Responsible AI: From Principle to Practice! Mehrnoosh Sameki Principal PM Manager @Microsoft 14:55 Mountbatten (6th Fl.) Session Graphs Graph Learning at the Scale of Modern Data Warehouses Subramanya Dulloor Founding Engineer @Kumo.ai 16:10 Mountbatten (6th Fl.) Session python Simplifying Real-Time ML Pipelines with Quix Streams: An Open Source Python Library for ML Engineers Tomáš Neubauer CTO & Co-Founder @Quix