The Feast Podcast - MLOps Open Source Feature Stores Feast OSS
-
- Technology
A podcast that centers around the Open Source Feature Store Feast. In this series, the creator of Feast Willem Pienaar and the founder of the MLOps Community Demetrios Brinkmann join forces to interview Feast users and each other on the current state of Feast, surprises to implementing a feature store, specific Feast use cases, and Machine Learning Engineering wisdom in general.
Feast has become the goto feature store for large companies such as Zulily, Gojek and Cookpad and is also the main feature store being used with the Machine Learning tooling Kubeflow.
-
Thinking Beyond the Feature Store and Data Warehouse With Miles Adkins of Snowflake
Miles is a Partner Sales Engineer at Snowflake and leads the technical go-to-market and joint product integration strategy for Snowflake's machine learning and data science partners. Miles has spent the entirety of his career in data science and sales engineering positions and specializes in financial markets and technology industries.
With the rise of data platforms that now can enable data sharing, and the ability for feature stores to create data pipelines as code, the direction is clear that the "Feature Cloud" concept is heading towards us rapidly. What will this mean for data providers in how they structure and make their data available? Where in the stack will be the processes that manage data provider SLAs. Who will vote on changes to feature definitions? These are just some of the questions that we will need to answer for shared data to power production machine learning use cases.
Want to see more about how Feast and Snowflake play well together? Check out this blog post.
If you would like to share your story with us about how you are using Feast, get in touch!
For more info about feature stores check out the Tecton blog or have a look at the Feast website
Join the Tecton/Feast slack channel to speak with like-minded engineers.
Connect with Willem on LinkedIn
Connect with Miles on LinkedIn
Connect with Demetrios on LinkedIn
Follow Feast on Twitter
Check out Tecton on Twitter -
Keeping Feast Simple with Danny Chiao // Feast Feature Store
Danny Chiao is an engineering lead at Tecton/Feast Inc working on building a next-generation feature store. Previously, Danny was a technical lead at Google working on end to end machine learning problems within Google Workspace, helping build privacy-aware ML platforms / data pipelines and working with research and product teams to deliver large-scale ML powered enterprise functionality. Danny holds a Bachelor’s degree in Computer Science from MIT.
In this session Danny tells Demetrios Brinkmann from the MLOps Community how he works with the Feast Community to drive design decisions. He gives some hot takes on how working at Google is much different than working for a startup, especially in the Machine Learning and Feature Store space.
The current Python Centric Feast meets you where you are, whether thats a small team of 2 data scientists, or a platform team at Twitter. The Feast ecosystem is built with scalability and performance capabilities in mind. Danny was conscientious about how with the Python centric Feast approach it is not throwing the kitchen sink at uses from the start. The idea now is to allow teams to add as much complexity as needed once up and running with Feast.
If you would like to share your story with us about how you are using Feast, get in touch!
For more info about feature stores check out the Tecton blog or have a look at the Feast website
Join the Tecton/Feast slack channel to speak with like-minded engineers.
Connect with Demetrios On LinkedIn
Connect with Danny on LinkedIn
Follow Feast on Twitter
Check out Tecton on Twitter -
Evaluation Store + Feature Store
Aparna Dhinakaran spoke with Demetrios and Willem about the 3 tools you need to do ML - an evaluation store, a feature store and a model store. Before getting into the full conversation about the integration between Feast and Arize AI, the team talked about what exactly an evaluation store is and why its needed.
Aparna's article about the three tools you need to do ML - https://towardsdatascience.com/the-only-3-ml-tools-you-need-1aa750778d33
The Feast and Arize integration blog - https://arize.com/feast-and-arize-supercharge-feature-management-and-model-monitoring-for-mlops/
If you would like to share your story with us about how you are using Feast, get in touch!
For more info about feature stores check out the Tecton blog or have a look at the Feast website
Join the Tecton/Feast slack channel to speak with like-minded engineers.
Connect with Aparna on LinkedIn or Twitter
Connect with WIllem On LinkedIn
Connect with Demetrios On LinkedIn
Follow Feast on Twitter
Check out Tecton on Twitter -
Feast's Vision For the Future
In this episode Jay Parthasarthy sat down to talk with Demetrios Brinkmann about the vision for Feast and what a Product Manager does for the Feast ecosystem.
If you would like to share your story with us about how you are using Feast, get in touch!
For more info about feature stores check out the Tecton blog or have a look at the Feast website
Join the Tecton/Feast slack channel to speak with like-minded engineers.
Connect with Jay on LinkedIn or Twitter
Connect with Demetrios On LinkedIn
Follow Feast on Twitter
Check out Tecton on Twitter -
MLOps Platform Teams
Oleksii Moskalenko joins Demetrios Brinkmann to talk about how the team at Gojek is using Feast. Some of the main work Oleksii is doing is on the platform team. This episode centers around what his team focuses on when enabling data scientist, and what some of the main use ML cases are at Gojek.
Oleksii is currently an Engineering Lead at Gojek's Data Science Platform, leading the Feature Store development. Prior to that he was Algorithm Engineer working on productionizing deep learning (computational photography) models at Wix. Moskalenko's main focus is helping to convert academia's accomplishments in Data Science into industrial software solving real-world problems. Overall he has 15 years experience in software engineering with the last 6 years focusing on Data Engineering & productionizing Machine Learning.
If you would like to share your story with us about how you are using Feast, get in touch!
For more info about feature stores check out the Tecton blog or have a look at the Feast website
Join the Tecton/Feast slack channel to speak with like-minded engineers.
Connect with Demetrios On LinkedIn
Connect with Oleksii On Linkedin
Follow Feast on Twitter
Check out Tecton on Twitter -
Kubeflow + FEAST With David Aronchick Co-creator of Kubeflow
David Aronchick has worked for tech giants such as Microsoft, Amazon and Google. He is most notably known for his role in creating the open source ML tool Kubeflow. Yes Kubeflow. In this conversation Willem, Demetrios and David talk about challenges the current MLOps ecosystem faces, where the market is going and the Feast + Kubeflow dynamics.
If you would like to share your story with us about how you are using Feast, get in touch!
For more info about feature stores check out the Tecton blog or have a look at the Feast website
Join the Tecton/Feast slack channel to speak with like-minded engineers.
Connect with David On LinkedIn or twitter
Connect with WIllem On LinkedIn
Connect with Demetrios On LinkedIn
Follow Feast on Twitter
Check out Tecton on Twitter