Can we have a better feature store? a talk by Artem Zhukov

Saturday 16 September 15:40 (30 minutes)

__floor__

In my talk I will cover the concept of a feature store in Data Engineering, its key properties, and the available solutions.

I will discuss benefits of developing your own feature store for your team’s growth and for meeting specific project requirements, adapt to specific team workflows, and better integrate with other software.

In the end I will review the model for a lightweight feature store, considering some notable design decisions including prioritizing simplicity in design and extensibility, using Delta format and putting an emphasis on features, not tables.

What do you need to know to enjoy this talk

Python level

Medium knowledge: You use frameworks and third-party libraries.

About the topic

You used or did it just a few times.

Artem Zhukov

zhukovgreen

I am a software and data engineer who started a professional IT career in 2016. Within data engineering, I have relatively good experience working with Apache Spark, Delta, and Databricks. I also love and use functional programming, Python's mypy, software architectures, distributed systems, and Python's asyncio.

I am writing tests before the implementation almost every time. And I am terrible at quick and dirty tasks. My last three projects are used by many people and keep evolving even if I dropped maintaining them.

I had various and very different experience with the CI/CD systems and clouds, building and maintaining various on-prem systems as well as systems built on top of AWS.