Supporting Diverse ML Systems at Netflix

Netflix uses data science and machine learning across all facets of the company, powering a wide range of business applications. The Metaflow machine learning platform at Netflix provides an entire user-centric ecosystem of abstractions and integrations that allow practitioners to tackle a diverse set of business problems.

In this talk, we will first introduce Metaflow, an Open Source Software, and how it accelerates the work of ML practitioners by providing simple and consistent abstractions over core properties of ML pipelines. We will show how components such as data, orchestration, hosting and others can be simply combined to iterate quickly from design to production over a wide range of complex data science and ML workflows. 

Through the talk we will highlight use cases to demonstrate the breadth of ML applications developed on Metaflow, and we will point out the user-design, technical and operational principles that helped us scale to hundreds of users and several hundreds of workflows while maintaining a small team with a focus on customer service and interactions.

The presentation will cover: 

  • How to build foundational components that can be combined to create novel ML applications across diverse use cases.
  • How to design user-centric systems that cater to a wide range of ML practitioners.
  • Technical and operational lessons about scaling and maintaining a large platform with a small team. 

From the same track

Session

Optimizing Search at Uber Eats

Uber has an in-house search engine called Search In Action (SIA). As the backbone behind the feed and search capabilities of Uber's Delivery business, SIA plays a crucial role in expanding selection seamlessly for customers which is a strategic advantage to the business.

Speaker image - Janani Narayanan

Janani Narayanan

Senior Staff Engineer @Uber

Speaker image - Karthik Ramasamy

Karthik Ramasamy

Senior Staff Software Engineer@Uber

Session

Slack's AI-Powered, Hybrid Approach for Large-Scale Migration from Enzyme to React Testing Library

With the Enzyme test framework no longer supporting React 18, migrating to React Testing Library (RTL) became imperative.

Speaker image - Sergii Gorbachov

Sergii Gorbachov

Senior Software Engineer @Slack