Building during my leave

I am spending part of my maternity leave building a data platform.

Not because I have to — I am on leave, I have to do nothing at all — but because it genuinely makes me happy.

I love my job as a data engineer. I love the power of a well-designed enterprise platform. But somewhere along the way I noticed something that has quietly kept bothering me.

Engineers in our field are increasingly trained on one type of tool: the big, expensive, fully managed SaaS kind. And that is fine, those tools can be fantastic. But it means fewer and fewer people know how to build something from scratch. How to connect several open-source tools and end up with something production-ready. How to look at a small company's data problem and reach for the solution that fits, rather than the solution that is familiar.

That skill — open-source system integration, deliberately building your own stack — is becoming rarer. And I think it is worth pushing back against that a little.

So Stan and I are building a small open-source data platform (datavloot) as a starting point for what that looks like in practice. DuckDB, dlt, dbt, Dagster.

Built to be understood, not only to be operated.

Will it change the industry? Probably not. Is it a new hyperscaler? Absolutely not — it is a hobby and a passion, and if the end result is simply a good ride and some newly gained knowledge, that is a completely successful outcome.

Follow along if this is your kind of thing.