yiksanchan

I create a few more and list them on opensea for sale :)

https://opensea.io/collection/sillycon-valley

ah gotcha

Thanks for the kind words! Anything I can do to resolve your issue? Is there a way to reproduce?

I visualize them in https://misbrands.vercel.app/

Code: https://github.com/yiksanchan/misbrands

Check it out!

Thanks for sharing!

This SQL-intensive actually reminds me of the feature store solution by Splice Machine (https://splicemachine.com/press-releases/splice-machine-laun...). They implement a HTAP database and use that in a similar as you use PL/SQL. I had a writeup on this as well (https://yiksanchan.com/posts/splice-machine-feature-store)

Sorry I miss that

I suggest checking out the link from @willempienaar. He is indeed the domain expert since he is building the open source feature store https://feast.dev/ :)

Thanks for your kind words!

> How do you handle historical backfill for new features?

Currently, our feature store doesn't come with inbuilt feature backfilling. In order to do that, some manual work needs to be done. We're working on a brand new version of feature store that hopefully addresses this need.

> Who is responsible for keeping the Flink jobs running: the data scientists, or do you have a separate streaming platform team?

We have a separate data infra team who is responsible for managing the YARN cluster for us.

Thank you for your kind words!

Hi NumberCruncher, thank you for your reply!

> do they have a feature store with built in time-travel functionality?

My feature store hasn't supported time-travel yet. But many SaaS implementations do, including Tecton, Hopswork, Splice Machine, etc. Open source feature stores haven't implemented this critical feature AFAIK.

> Why I didn't write about it? Because it was implemented in PL/SQL running on Oracle ExaData and in SAS. No one cares about feature stores implemented with tech like that.

Haha, I do. Actually I am thinking about implementing feature stores using some "old-school" DB technology. By the way, just curious since I've never used PL/SQL: Are you able to implement the time-travel functionality using pure SQL?

Thanks for the comments!

> This thing reads like it was written a few years ago.

Yah the technology here is nothing novel, Hive, Kafka, Flink, Redis are around for years. What I find missing in the internet is that people who have been doing this for years are not writing about this. Uber has done a relatively good job on publishing how they build Michelangelo, but still, not enough details for outsiders to replicate.

> it takes time for the lessons in this article to be learnt by a large enough amount of people that it becomes humdrum.

Maybe :)

> two clients who built feature stores - one with an open-ended goal of enabling people and one because they had some specific things they wanted to achieve.

Could you add more color to this part, what are their goals and what do they end up achieving? I didn't fully get it.