Works of... art (?) Software

Participating in the development of software is my day job, but since way before I was formally employed and to this day, I have also spent (too) much leisure time creating a lot of software and operating a handful of web services.

Much of the software I built is permissively licensed and, I believe, is sufficiently showcased in my GitHub and Tangled profiles. As for the rest, some of it is mentioned in posts on this website; others might only come up when you search for my name. Others even, were never mentioned online, despite reaching production status.

My list of projects is so long, and so full of things that are barely relevant these days, that enumerating feels like a daunting task. For example, throughout high school, I built a lot of software for semi-obscure calculators; years later, when I was already earning my bachelor’s, I built Clouttery and soon after, while working on my master’s degree, I developed an interest in the Lisbon Metro that led to a particular software-as-a-service. Those are all things you will discover by carefully searching the web… but you wouldn’t find out about my adventures in developing an internet-enabled curtain controller for my home, and I honestly don’t feel like writing much about it; I never opened its source to the public, and it is no longer relevant, in a world where Shelly does a much better job than my project ever did.

Allow me to go slightly meta for a second.

As Large Language Models evolve and so does the political and marketing discourse around them, I feel that software development is in a bit of a weird situation, where the activity/profession itself is changing as quickly and broadly as the many fields it once impressed its innovations on. Some might describe this situation as “tech” being a victim of itself, others would strongly disagree with the “victim” description. I suppose it depends on what you enjoy the most about software development, but I digress.

I am aware that software development evolved a lot over the decades, not just in the last few years. I am also aware that many other activities are feeling the power of LLMs in their daily workflows. Even then, I don’t think anyone in software development has ever felt that times are changing in the profession, as much as since the advent of code completion language models. It is a feeling that was once reserved for the fields that software was modernizing dramatically fast. Now, development (and cybersecurity, and…) is also dealing with extremely rapid technology-driven change, to the point where processes and approaches defined in the first quarter of a year might look a bit outdated by the last quarter.

I am not sure how good and relevant a list of non-AI-assisted personal projects will be in 5 or 10 years. Such a list is definitely still impressive today; hopefully, it would show that someone is able to hold their weight regardless of tooling. It should, hopefully, be a signal that someone’s a good candidate to bring some sense to the LLM-induced chaos, rein in quality issues, ask the right questions to the coding agents. However, I fear that in not too long, such an extensive portfolio predating “vibe coding,” will appear to be as relevant as someone’s accomplishments in DOS application development look today. That is, an indication of seniority, maybe a sign of great fitness for a very specific role, but little else, especially not in the way of adaptability to current tools and paradigms.

I strongly suspect that, within my lifetime, “hand-made software” (however that will end up being defined) will belong in the land of the creative arts, even more than it does today. I hope I’ll be able to find solace in my foresight to create a page for Software in the “works of art” section of my website, in the first half of 2026.