About
I'm Hotaviano Melo, an engineering leader based in Brazil. I lead a senior architecture team building software with AI at consumer scale (4M MAU): setting the patterns and platforms a product runs on, and making agentic engineering the way we build.
My work sits at one intersection: AI-powered software engineering, done agentic and eval-driven, applied to real systems at scale. That means software architecture, distributed and data-intensive systems, and the discipline that matters most once AI writes the code: finding the right problem and judging the output.
The loop I work and write by is simple: find the right problem, validate it's worth it, ship it, and measure the outcome. I care about results that are measured, not vibed, and about assets that hold up over opinions that don't.
This site is where I document what actually works, and what doesn't, as I build. No prompt tricks, no demos, no hype. If there's no build behind a post, it stays an honest note or it doesn't ship. I write in English so the work travels.