Connecting the Dots

Software engineering was my sixth choice.

In Portugal, you apply to university with a ranked list of six options. Biology was my first. My grades were not high enough, and I ended up in the program with the lowest entry bar that year: software engineering.

I did not want to be there.

The first year was brutal. Algorithms, Boolean algebra, logic structures. Everything felt foreign. I came from a mindset of observation and investigation, the kind of thinking where you watch the world and try to understand why things behave the way they do. Software asked me to think differently, and I could not make the switch. I failed exercises that seemed obvious to my classmates.

Then a professor named M. Neves, who held a PhD in Robotics and Engineering, invited me into a research project. It changed everything. Not because the work was extraordinary, but because it showed me software could touch the real world. The investigator in me found a new tool.

I stopped comparing where I was to where I thought I should be. I accepted this was my field, and I chose to go deep into it.

What curiosity built

My first real job was at a company more than twenty-five years in insurance software. As a junior, I wrote small pieces of code. But what I actually learned came from observation: how systems stayed alive in production, how deployments worked, how software lived on infrastructure someone other than the developer had to care for.

Years later, while leading development elsewhere, a CI/CD pipeline kept pulling my attention. I used it every day and had no idea how it worked underneath. I grabbed an old computer, installed Git and Jenkins, and started figuring it out. That was the start of my home lab.

It grew because life gave me real problems. No static IP, so I learned dynamic DNS and VPNs. A disk failed, so I learned storage, backup, and automation. Over time, it became a five-node cluster. None of it was in my job description. I learned it because I needed it.

Then Home Assistant brought sensors into the house: temperature, doors, light. Software responding to the physical world again. It felt like what I would have done in biology, with different tools.

My two lives, professional software and personal infrastructure, were running in parallel.

Where it meets the work

When I joined Powerdot, I had more than ten years building software. I wanted to work where code connects to the physical world.

Powerdot manages electric vehicle chargers across Europe. Critical infrastructure exposed to weather and wear. The architecture is IoT at scale: receive inputs from the field, process them, send actions back. More complex than my home lab, but the concepts underneath are the same.

I did not join just for the software. I joined because it felt like the place where my professional life and personal obsession finally overlapped.

Then I was given the responsibility to lead AI enablement across the company. Working with the steering committee, defining how over three hundred people would adopt AI across operations, product, and engineering.

What made me the right person was not AI expertise. It was years of self-hosting, open source instinct, and a belief in infrastructure independence. Powerdot cannot depend on a single AI lab or vendor that changes pricing overnight. Resilience means controlling your own stack. That is exactly what my home lab taught me.

My hobby became my job.

Where this leads

I wrote about Amor Fati a few weeks ago. Loving your fate. Embracing where you are instead of resenting where you think you should be.

Looking back, every stage carried something forward. None of it was planned. What I did was accept where I was and go deep into it.

The vision for what comes next is simple. I want the work I do to compound the same way the home lab did. Build the foundation yourself. Own the dependencies. Let the surface change, but keep the underlying structure stable.

The strategy is equally simple. Follow the curiosities that pull into the margins, especially when they look unrelated to the main path. Invest in skills before they are needed. Place the dots without knowing how they connect.

That is what I am doing now with AI at scale. Not because it is the trend. Because it demands exactly what everything else has demanded: build the infrastructure yourself, own your data, and think in decades instead of quarters.

The dots only connect looking back. While you are living it, all you can do is keep placing them.