Independent Full-stack AI Engineer
I build full-stack AI-native software.
I build AI-native applications and platforms for government and enterprise operations.
AI projects don't fail because the model is wrong. They fail because the AI is grafted onto software instead of designed into it. The model is the easy part. Everything around it is the work.
Companies and organizations that trust my work
Everyone is shipping AI. Few are shipping AI that gets used.
In any company running operational software — internal operations, back-office, and critical platforms — adding AI has become table stakes. The model works, the code is fine, but most features die silently because no one uses them.
The problem is rarely a single technical issue. It's that architecture decisions — RAG, context engineering, guardrails, fallbacks — show up to the user as friction that breaks trust. Each architecture decision defines whether the tool gets used or ignored in production.
That intersection — between AI implementation and user experience — is where the system gets used or ignored. And that's where I work.
I build AI into the software that runs the business — not next to it.
The differentiator your product or operation needs isn't more AI. It's AI integrated tightly enough that people use it without thinking about it.
That's not won with a better prompt or a faster model. It's built in how the architecture is decided, how context is managed, and how the AI reaches the user.
Where I add the most value:
- →Building AI applications end-to-end — from use cases through architecture through interaction.
- →Auditing AI features that aren't being used, finding what broke trust.
- →Helping established organizations evolve their operational software toward AI-native architecture.
- →Fractional AI ownership: owning AI delivery so your team can focus on the rest of the product.
I work across web stacks (PHP/Laravel, Node/Express, Vue/Nuxt, React/Next, TypeScript, Postgres) and AI tooling (Claude, GPT, Python, embeddings, vector and graph stores, MCP) — building RAG pipelines, agentic systems, workflow orchestration, memory architectures, and operational automations. The work also covers prompt, context, and harness engineering, guardrails that keep systems reliable in production, and interfaces that work with AI uncertainty instead of hiding it. It isn't about which framework — it's about which decisions to make in any of them.
Most AI engineers ship features that don't land.
Most AI engineers deliver features that technically work but don't land in real use. Architecture decisions have to hold up against what people actually do with the system — not be improvised on top of a model.
20 years of software architecture. 8 years of UX as an explicit discipline. The last few building AI in production. That combination is what makes the difference between AI that ships and AI that gets used.
Industrial B2B Support Tickets
Conceptual design · Technical demonstration
Designed and built two AI agent systems, one for operators, one for technicians, connected by a structured context object. The AI routes, pre-fills, and delivers diagnostics with explicit confidence levels to the user.
50% reduction in resolution time (projected) · 75–88% of ticket context pre-filled automatically (projected)
TÜV NORD Argentina · Sistema Cóndor
System design · Enterprise platform · 2 years in production
Designed and built the complete platform that centralized TÜV NORD Argentina's operations into a single system. Its capabilities include an AI layer that generates editable pre-proposals, preserving user control without losing automation efficiency.
80% less administrative overhead · 100% of operations centralized
Mi Caja — Government Platform
System architecture · Government · 100k+ active users
Designed and built a platform that replaced in-person bureaucratic processes with digital workflows, at scale. The foundation from which I now build robust software with integrated AI.
70% reduction in processing times · 100,000+ active users in production
Let's talk about what you're building.
Available for full builds with integrated AI, audits of features that aren't landing, or help evolving your operational software toward AI-native architecture. I'll respond in under 24 hours.