Automating language migration at scale: tools, tradeoffs, and the role of AI

Migrating tens of millions of lines of Java at Meta required more than incremental tooling. Eve Matthaey, software engineer at Meta and lead of the Kotlin conversion project, breaks down how her team built custom refactoring infrastructure to run a full-scale, automated language migration across an actively developed codebase without stalling developer velocity. She covers how metaprogramming, static analysis, and runtime data collection were combined to solve two of the hardest problems in large-scale migration, null safety and dependency graph complexity, and the tradeoffs that kept developer toil and risk low at Meta’s scale.