The longer version.

I was not the kid who always knew he would become a software engineer. I did not grow up with a neat plan that connected every interest and made the ending obvious. For a long time, I was simply interested in why some things worked and others did not — businesses, decisions, incentives, people.

That curiosity led me to economics at PUCPR, here in Curitiba. Economics gave me a language for questions I already liked asking: what does something really cost, what happens when incentives point in the wrong direction, and why do sensible decisions sometimes produce bad systems?

I liked the rigor of economics and the suspicion it teaches you to have toward confident numbers. A spreadsheet can make almost anything look precise. It can also hide the fact that the process producing those numbers is held together by messages, memory and somebody staying late.

As I began working with businesses, I noticed that many financial problems were not really financial. Money was being lost because information arrived late, ownership was unclear, a handoff depended on one person, or a decision was made with half the context. The spreadsheet could describe the damage. It could not repair the operation behind it.

So I started learning to build software. Not as a dramatic reinvention, and not because I had stopped caring about economics. I was impatient with watching the same avoidable problems happen again. Code gave me a way to change the process instead of explaining, one more time, why the process had failed.

Around 2020, I began working independently. Freelancing became my real engineering education. For the next six years, I worked with B2B and B2C clients across different industries, usually beginning with something much less organized than a product brief. A client would arrive with a slow operation, a messy sales process, an e-commerce problem, a report nobody trusted or a task that consumed far too much of somebody's week.

I had to learn the business before I could build the software. That meant asking uncomfortable questions, separating the symptom from the constraint and deciding what not to automate. It also meant moving through the whole stack: interface, backend, database, integrations, deployment and the less glamorous work of making sure somebody else could operate the system after I left.

One early client process took two or three days of back-and-forth to complete. I rebuilt it as a structured web workflow and brought the turnaround below four hours. I remember that project because it clarified what clients were actually paying me for. They did not care which framework I used. They cared that two days of friction had disappeared.

B2C work taught me to respect attention. People do not read the interface with the patience of the person who designed it. B2B work taught me to respect consequences. A small ambiguity in a workflow can become an expensive operational habit when an entire team repeats it every day. Both taught me that good software needs to explain itself and then get out of the way.

Freelancing also made me comfortable with incomplete information. Most useful problems do not arrive with acceptance criteria. They arrive through a conversation in which the client describes three different problems, contradicts themselves twice and eventually says one sentence that reveals the real issue. I learned to listen for that sentence.

In 2023, I tried turning those instincts into a product. SkytOffer helped digital launchers diagnose weak offers before spending more money on traffic. Users brought a promise, an audience and an offer structure; the system helped expose gaps in positioning, perceived value, urgency and risk reversal.

SkytOffer was my first serious lesson in the difference between consulting and product. A consultant can explain the missing context. A product has to survive without you in the room. Prompts that looked clever in a demo broke against real users. Advice that sounded intelligent was useless if it did not change the next decision.

In one case, the diagnosis showed that traffic was not the problem. The promise was too broad and the offer did not answer the buyer's main objections. After both were reworked, conversion improved by 28%. SkytOffer never became a large business, but it gave me the foundation for the product I built next.

AlturionX was not a separate invention so much as SkytOffer rebuilt into a broader B2B product: AI-assisted internal tools and workflow systems. I was the technical founder, which in a small operation is a grand title for doing whatever the week requires. Product direction, frontend, backend, database design, infrastructure, LLM integrations, sales calls and support all belonged to the same job.

The hardest decisions were rarely technical. With a small number of clients, every client matters. A custom request can feel urgent and still be the wrong thing to turn into permanent product complexity. I learned to say no more carefully, to price the maintenance hidden inside a yes, and to distinguish what one customer wanted from what the product needed to become.

One acquisition workflow turned scattered lead context into a structured qualification and offer-positioning process. It contributed to a 38% increase in new client conversion in one operation. The result did not come from giving an AI model more authority. It came from building a disciplined workflow around a narrow model task.

I kept consulting while building the product. At Inova Recurso, I worked on proposal and operational workflows. At Metabio, I helped turn fragmented campaign data into a reusable reporting process. Moving between my own product and other businesses kept me honest. The industry changed, but the useful question remained the same: where is the work actually getting stuck?

Somewhere in the middle of those years, life became larger than work. I got engaged.

I will not pretend that getting engaged transformed me into a different person overnight. It did make the future less abstract. Decisions stopped being only about what I wanted to build next and started including what kind of life we wanted to build together. Ambition still matters to me. So does being present, keeping promises and becoming someone another person can safely plan a life with.

Building companies teaches you to optimize almost everything. A relationship reminds you that not everything valuable should be optimized. Some evenings are allowed to be slow. Some decisions need patience rather than a framework. Some parts of a good life are built through repetition: showing up, listening properly and making room for somebody else's version of the future.

Curitiba is home. Most late afternoons, I end up at Parque Barigui with Mick, my small white dog and least impressed critic. We walk around the lake while the sun goes down. Many difficult AlturionX decisions became simpler somewhere along that path, usually after I had stopped trying to solve them at the desk.

Mick does not care about software, revenue or whether a launch went well. That is probably healthy. The walk creates a boundary between the work and the rest of my life — my relationship, the people I love, the city I chose to stay in and the person I am when there is nothing left to ship that day.

Eventually, I sold AlturionX's product assets: the codebase, the operating platforms and the client portfolio. The buyer remains private by preference. By then, the product had served 16 high-ticket B2B clients. The sale taught me that buyers value many of the things builders postpone because they look boring: retention, documentation, repeatable operations and a product that does not collapse when its original builder steps away.

The sale did not feel like an ending. It felt like evidence that I could take an unclear problem, build a product around it, earn trust from real customers and leave behind something another person believed was worth continuing.

Today, I still build from the same basic conviction that pulled me away from spreadsheets and toward software. The best systems make a difficult piece of work clearer. They use deterministic code where being wrong is expensive, AI where it earns its place and infrastructure small enough that a real team can understand it.

I am now looking for the next team where I can bring all of that together: six years inside B2B and B2C problems, full-stack implementation, founder-level ownership and the judgment to know when the most impressive solution is not the most useful one.

I still do not have a neat plan connecting every chapter. I have a life I am building, a person I am building it with, a dog waiting for his walk and a long list of systems I would like to make less frustrating. That feels like enough direction for now.