all posts
Data 10 min read

Scala 3 in anger: what changed for our data team

We migrated our core data libraries to Scala 3 a year ago. This is the honest retrospective — what helped, what cost us, and what we still avoid.

Earned their keep

  • Given/using clauses made our implicit-heavy code finally readable to newcomers.
  • Enums replaced a pile of sealed-trait boilerplate.
  • The new error messages are dramatically better, which matters more than any feature.

The migration itself cost about three engineer-weeks of cleanup. For a codebase this central, that paid for itself within the quarter.

// KEEP READING
Data

The petabyte pipeline that stays debuggable

Scale is easy to add and hard to keep legible. The discipline that kept our Spark pipelines understandable as they grew past a petabyte a day.