GitHub uses AI to migrate 800,000 lines of code from TypeScript to Rust

TVS News AI, editorially supervised, presents GitHub uses AI to migrate 800,000 lines of code from TypeScript to Rust.

Big code migrations hit GitHub itself this time: According to Golem.de, the company migrated around 800,000 lines of code from TypeScript and Node.js to Rust — with the support of artificial intelligence.

For developers, the news is more than a tech tidbit. It shows the scale at which AI-assisted language switches are realistic today — and where humans still need to step in.

What the AI could and could not deliver in this project, you can read here.

What was migrated from TypeScript and Node.js to Rust?

Around 800,000 lines of code previously written in TypeScript and running on Node.js were ported to Rust, according to Golem.de. Rust has built a reputation over the years for its memory safety, performance, and strong concurrency — especially in the backend realm, where every millisecond and every megabyte count.

What role did the AI play specifically?

A migration of this scale would classically involve enormous manual effort: reading each file, translating, testing, reviewing. This is precisely where AI came in, according to Golem.de. The report states:

Although the migration project by GitHub was economically feasible with the help of AI, human oversight remained essential.

The specific tools or models used and how much time was actually saved are not disclosed in the report. What is clear, however, is that the AI reduced the effort enough to make the project economically viable.

Why did human oversight remain essential?

Anyone who has reviewed AI-generated code knows the problem: what looks correct in the source language can be subtly wrong in the target language. With Rust, this is especially tricky — and especially helpful at the same time. The compiler is strict and flags many errors early. Even so, you need eyes that understand why the AI chose a particular translation — especially when moving from a language with garbage collection like JavaScript to one with manual memory management.

What does this mean for you as a developer?

GitHub runs a platform hosting millions of repositories. When 800,000 lines of code are migrated there, it sends a signal: AI-assisted language migrations have arrived in industry, even at scale. For your own work, this means: a migration that used to be a year-long project could become significantly smaller in the future — provided the review team is large enough and the AI tools are robust enough for the target language.