Twitter’s Feature Store Journey

apply(meetup) - Feb '22 - 10 minutes

A Feature Store is an essential piece of a production ML system. Twitter’s journey of building Feature Stores began several years ago. Since then, we have gone through multiple iterations of our Feature Store to facilitate creating, organizing, sharing and accessing ML features in production. In this talk we will touch on the key parts of this journey, the move from a virtual to a managed store, and our decision to adopt Feast. We will share our reasoning behind design decisions, the challenges we encountered, and the lessons we learned.


Tzvetelina Tzeneva

Software Engineer

Twitter

Tzvetelina is a Staff ML Engineer in the ML Platform Organization. She is the tech lead of the ML Feature Management team and has worked on building two generations of Feature Stores at Twitter. Prior to Twitter, Tzvetelina worked at TellApart where she built distributed systems infrastructure, data platform and data pipelines for the large scale application of direct response advertising.

Outside of work, Tzvetelina is a recreational Aerial Silks instructor.