I turn ideas into production-ready digital products.
I combine product thinking, full-stack engineering and cloud architecture, applying AI where it creates real value.
8 years across software engineering, architecture and technical leadership. TypeScript, AWS and applied AI.
From transactional TypeScript and AWS platforms to AI products for web, macOS and iOS.
Three ways of working together. All three end the same way: something real running in production.
I work with founders, product teams and engineering leaders who need senior ownership from an unclear problem to a reliable production system.
I design and build products that incorporate LLMs, voice, agents, RAG or intelligent automation, from prototype to a reliable system in production.
You get an AI product that works for real users: with guardrails, measured cost and latency, and deterministic fallbacks when AI is not the answer.
I develop products and features end to end, from problem definition through delivery, measurement and iteration.
A senior engineer who owns the outcome, not the ticket: discovery, implementation, deployment and iteration in a single profile.
I design and evolve complex systems, and support teams hands-on when they need technical direction without separating architecture from implementation.
Decisions that hold up in production: clear boundaries, safer integrations and a team that ships with more confidence.
Real products in production: my own AI-enabled apps, client work and an enterprise product case.
AI product — concept, design & engineering · Open source (MIT)
Conventional dictation fails on technical language, Spanish–English switching and privacy. ChacharApp is the product: local voice dictation for macOS — hold a key, speak, and the corrected text lands at your cursor, typically in about one second on Apple Silicon, with nothing leaving the Mac.
Indie iOS product · App Store
Shift-work calendar for iOS: photograph your rota or pick the PDF and get a colour-coded month in under a minute. AI only where it adds value — deterministic extraction when the PDF already has text, no accounts and local-first storage.
Software Architect & Tech Lead · PostNL Digital Lab · 2022–2025
Three years inside PostNL's Digital Lab turning prototypes and ambiguous product requirements into a transactional checkout product connecting e-commerce platforms, shipping and payments — as architect, tech lead and hands-on engineer.
How I run agents today · hooks, specs, subagents
An LLM is non-deterministic, so the critical rules have to live outside the agent, in code the agent cannot rewrite.
The same loop on every project, with or without AI.
Clarify users, business, constraints, risks and the success criteria before choosing any technology.
Define flows, boundaries, experience, architecture and the critical decisions.
Ship complete increments that cut through interface, backend, data and infrastructure.
Measure quality, behaviour, latency, cost, security and how users respond.
Watch production, correct assumptions and evolve the product.
I also build with code agents: executable specs and controlled environments (a harness) that increase speed without losing quality or judgement. Agents are a tool — the responsibility for architecture, code and production stays with me.
Software delivery, AI product ownership and production architecture in one hands-on profile.
I build complete, maintainable software across the frontend, backend, data and integration layers.
I shape, build and iterate AI-enabled products around real user needs and measurable outcomes.
I design reliable systems that can evolve as the product, team and usage grow.
TypeScript · React · Next.js · Node.js · NestJS · Swift · SwiftUI · PostgreSQL · Prisma · AWS · Terraform
The full detail lives in the resume — this is the shape of the journey.
I build digital products for clients and develop my own AI-enabled products, including ChacharApp (local dictation for macOS) and Salturno (shift calendar for iOS).
Led the architecture and hands-on delivery of Fast Checkout: from ambiguous product requirements to a transactional e-commerce product on AWS, coordinating several teams and third parties.
End-to-end systems from the first line of code — including GICA, the wildfire response management system used daily by 2,200+ field workers.
Tell me what you are building, what is blocking the team and what outcome you need.
Essays on product, engineering, AI and building a career with independent judgment.
An LLM is non-deterministic, so the critical rules have to live outside the agent, in code the agent cannot rewrite.
ReadCode is now mass-produced, and the reader spending the most time on it is an AI agent. I built an entire app with one, in a language I don't master, and the result forced me to rethink the old rule that code shouldn't be commented.
ReadJust having faith that things will work out isn't a strategy.
Read