Design engineering, after 18 years across agencies, product companies, and international financial institutions.

Designer roots

I started in Brazil in 2005, first in graphic design and then on the web, at a time when the line between designer and developer was both rigid and impractical. I was designing interfaces, handing them over, and watching the quality change somewhere between the layout and the browser.

Learning HTML and CSS was not a career rebrand. It was a way to protect the work. I wanted the interface that shipped to keep the intention, rhythm, and detail of the design.

The frontend decision

By 2013, the middle ground had become too vague. The market still wanted specialists, and I had to choose whether to return to design as a separate discipline or keep moving towards frontend with intent.

I chose frontend because it kept design and implementation in the same conversation. That path moved through product UI, JavaScript, React, TypeScript, reusable interface patterns, and eventually the design systems work that made the connection explicit.

International product work

From 2015 to 2025 I worked in Ireland, inside international financial and government clients where a UI decision had to survive regulatory review, distributed teams, and long-lived codebases.

Those environments included Bank of America, Bank of Ireland, Hertz, Walmart, Brazil's Ministry of Health, Brazil's Supreme Court, AnPost, and AA Insurance. What I took from each was the same: write the component name or the token rule so the next engineer can follow it without asking.

Community and writing

Choco la Design was an important part of that path: a place to write, share, and stay close to the design community while I was moving deeper into implementation. It also kept the original question alive: how do designers and developers work better together without losing the quality of either discipline?

Current focus

Today the work centres on design systems, product frontend, and AI agent workflows. AI is interesting to me when it makes a spec review, a code review, or a research summary faster to check. Not when it generates confident output nobody can verify.

The short version: find where the Figma file and the codebase disagree, fix it at the token or component level, and write down why so the next person does not have to ask.