Technology is becoming more intelligent. That doesn’t automatically make products easier to understand. I design products where trust cannot be assumed. From AI-powered experiences to complex enterprise systems, I help teams transform uncertainty into products people can understand, inspect, correct, and confidently act on.
13+ years designing decision-critical products across GovTech, fintech, enterprise SaaS, real estate, and AI.
- Brasília, Brazil · UTC−3
- Working remotely across the Americas and Europe
- Open to relocation
Clients and partners include Nortal, Capgemini, Ília Digital, Markerr, Santander, TCU, Vinci Partners.
A few outcomes I’m particularly proud of.
40%
Revenue growth before acquisition
My work on Data Studio and Ask Markerr helped Markerr grow from data provider into product company — a repositioning that ended in its acquisition.
63%
faster audit reporting e-TCE
Led the UX of an AI-powered audit platform — redesigning the workflows behind the writing of audit instructions and reports, in one of the country's most skeptical institutions.
20+
Products and Initiatives
Designed across public services, finance, real estate, enterprise platforms, machine learning, and AI-assisted workflows. Twenty-plus products. Far more edge cases.
Every project shows its proof up front — a measured outcome or an honest label, before you click.
How I work
My process changes with the problem. The principles below usually do not.
When everything feels complicated, direction becomes the product.
Every complex product eventually reaches a point where nobody agrees on what the real problem is. That’s usually where I become useful. I help teams untangle systems, competing priorities, and technical constraints until the next decision becomes clear enough to move forward.
Design the failure states, not just the happy path
Intelligent and high-stakes products rarely behave perfectly. I design for uncertainty, partial success, correction, recovery, and the moments when users need to take control.
Happy paths make good demos. Recovery paths make trustworthy products.
Prototype close to reality
I use coded and AI-assisted prototypes when static screens cannot answer the important questions — especially around interaction, system behaviour, responsiveness, and feasibility.
The closer the prototype is to reality, the more useful the conversation becomes. Whenever possible, prototypes use the product’s real components, tokens, and interaction rules.
I started in advertising trying to understand how people make decisions. Later I became interested in the systems behind those decisions. Today I’m exploring how AI changes both. Different technologies. Surprisingly similar questions.
That’s what this portfolio is really about.
Explore how I think
Designing AI workflows people can understand, inspect, correct, and control.
- LLM and conversational interfaces
- Generative UI
- Human-in-the-loop workflows
- Verification and correction patterns
- AI failure and recovery states
In other words: bring me the problem that keeps getting described as “a little complicated.”
Off the clock, I go looking for focus in other forms:
You will usually find me practising kendo, reading about human behaviour and technology, planning a trip, playing tabletop games, or enthusiastically testing a tool I discovered five minutes ago.
Kendo teaches discipline. Zen teaches attention. Games teach systems. Travel reminds me that most assumptions are local.
Somehow, they all find their way back into the work.
