● Developer journal — Lisbon
I run a translation company. I also build the systems that run it.
M21Global has been translating since 2005. When generative AI started eating into our revenue, I used it to rebuild the company instead: the AI writes the code; I decide what it builds, and I answer for what ships. The real method, the real code, the failures included. Written here first; the business half goes to LinkedIn.
Writing
RSS ↗- 01/13 min
The Translation Research Is Public. Why Do We Find So Little of It Available in the Market?
Seven techniques from public research that separate high-quality AI translation from what the market sells: which ones the platforms shipped, which they left on the shelf, and how each one runs in my production engine.
llmtranslationengineering - 02/17 min
How My AI Translation Engine Runs, Day to Day
Is an AI translation engine easy to use and easy to maintain? A straight answer, built on how the system actually works: what the project manager does, what the engine does, and what keeping it running really takes.
llmtranslationengineering - 03/23 min
The Consensus Loop: AI Translation You Can Trust, With the Numbers to Prove It
How an LSP owner rebuilt his translation engine around a multi-model consensus loop: the business case, the metrics that matter and the design decisions to build the same.
llmtranslationengineering