I started as a developer over twenty years ago, grew into software architecture, and eight years in I added UX as an explicit discipline — not as a side interest, as a way of thinking about every system I designed.

For most of those years, the question that pulled me was the same: why do systems that technically work the same produce completely different experiences? Why do some teams adopt a tool with enthusiasm and others abandon it in two weeks? Why does the feature one team spent months building end up ignored by the users who supposedly needed it?

That pattern has been repeating across every kind of software for years. With AI, it got sharper.

The model works. The code is fine. But nobody engineered how AI reaches the user — the uncertainty states, the fallbacks, the moments where the system needs to hand control back to the human, the way the agent communicates what it knows and what it doesn't. That's not a technical problem in isolation. It's an architecture problem that manifests in real-world usage — engineering decisions made upstream.

In internal operations and back-office specifically, this is the dividing line right now. Everyone is shipping AI. Few are shipping AI that users actually adopt into their workflow. The difference rarely lives in the model or the prompt. It lives in decisions like: what does the agent see in each turn (context architecture). How does latency from a specific retrieval pattern shape whether the user trusts the answer. How are guardrails and fallbacks designed so the system fails gracefully instead of breaking trust silently. Every one of those is a technical decision that also defines the experience. There's no boundary.

That's where I work now.

I build AI-native applications and platforms for government and enterprise operations. AI integrated into the system, not added on top — where the technical decisions and the user-facing ones are inseparable, not because I sit between two disciplines but because in AI systems they're the same decision.

What that combination produces is specific: AI features that get used instead of just shipping, technical decisions sustained by what users actually need, and a system that gets used the way it was intended.

I work directly, without heavy structures. A single point of contact who understands the full problem — architecture, engineering, and how the system reaches the user — without you having to coordinate between three different people. I don't deliver decks with abstract recommendations. I deliver documented system designs, evaluation frameworks, architecture decisions, and functional implementations your team can build on starting the next day.

Production systems for TÜV NORD Argentina, Ministerio de Seguridad Nacional, and others. I work independently from Buenos Aires, with teams across Latin America, the US, and Europe.

The first conversation is free. If it makes sense to work together, we'll know on that call.