Software Engineer · Málaga, Spain
I turn business problems into working software.
A process still running on spreadsheets. A system nobody understands any more. An idea that keeps not getting built. I work out what is actually needed, then build it — with AI and agents where they earn their place, and plain software where they do not. Eight years of this, across hospitals, agriculture and retail.
What I do
Working out what to actually build
Most wasted software money goes on the wrong thing, specified badly. I start with the people who have the problem — not with a technology — and write down what the solution has to do before anyone opens an editor. Eight years of doing this from both sides of the handover.
Digitalising work that runs on spreadsheets
A process that lives in email threads, a shared drive and somebody’s memory, turned into a system people will actually use. Usually the hard part is not the software. It is understanding the process well enough that the software does not make it worse.
AI and agents, pointed at real work
The model was never the hard part. Connecting an agent to the systems, data and approval rules a business already has — so it does something useful and can be trusted with it — is the hard part. I build the ones that survive contact with production, and I have taught teams to keep them running.
Making systems talk to each other
Integration between what you have and what you are buying: APIs, data pipelines, and databases that predate everyone in the building. Hospital record systems, agricultural prediction engines, retail data platforms — the plumbing is always where the value gets stuck.
Selected work
Software Analyst · REWE digital
2025 – present · Málaga
A data-hub team inside one of Europe's largest food retailers, feeding the shelf-layout planning that decides what sits where in a store. I was hired as a business engineer and work closer to a founder engineer: running the stakeholder interface, writing the specifications the team builds against, and building the tooling — monitoring, testing, AI agents — that shortens the distance between a business question and something shipped.
Kotlin · Java · Python · Kafka · PostgreSQL · Db2 · Snowflake · GCP
Software Engineer · Syngenta
2020 – 2023 · Basel
A research team's crop-prediction engine could only be used by asking them, and every request was answered by hand. As lead developer I built the serverless API that opened it up to business users inside the company and clients outside it — a documented contract, IAM-controlled access, structured forecasts. The manual request process ended, and the service ran in production for at least five years.
C# · .NET Core · AWS Lambda · API Gateway · IAM · REST
Software Engineer · Tekever
2018 – 2019 · Lisbon
A private hospital group in Portugal had self-service check-in kiosks that almost nobody used. As hands-on technical lead for a team of two, I took the software over from the outgoing supplier and rebuilt it around how patients and staff actually worked — integrated with hospital databases that predated the project. Self-service check-in success rose by 50%, the kiosks are still in production, and the work won the team the ticket-panel system that followed.
C# · .NET Core · React · MySQL · Docker · Microservices
Writing
The rules always end up in the code
Three times in ten years I have run into the same problem from a different direction: business rules that exist only inside software, and nobody left who can read them back out. What changed recently is not the problem. It is the cost of asking.
Your EHDS deadline is not 2029
The European Health Data Space gives healthcare providers until March 2029 to exchange patient summaries and ePrescriptions across the EU. Anyone treating that as the deadline has misread how long this kind of work takes.
About
I am a builder. I started as a backend developer — hospital systems, bikesharing, agricultural prediction APIs — and kept ending up in the room where the requirements got decided, because the interesting question was never how do we build this butwhat should exist at all. That eventually became the job.
So I do both halves. I sit with the people who have the problem, work out what the solution actually has to do, and then build it. Most places split those two jobs across two people and lose the meaning in between. The stack follows the problem: .NET, Java, Kotlin, Python, JavaScript, whatever the situation already runs on.
The AI part came from the same instinct. Rather than talk about what LLMs might one day do, I built things: agents that generate tests, produce documentation, and pull business rules out of code nobody wants to touch — then ran the workshops that got other people using them.
Healthcare is where I keep being drawn back. My first serious system was for a hospital group, and the problems there — old records systems, processes that cost patients their afternoon, data that cannot move between the people who need it — are the ones I find worth solving.
Portuguese, EU citizen, based in Málaga after stints in Lisbon, Basel and Beijing. I speak Portuguese, English and Spanish, and enough German to be polite about it.
Get in touch
If you have a problem you think software should solve, I would like to hear about it.
Send me a message and it lands in my inbox. No form-filling theatre — a name, an address to reply to, and whatever you want to say.