The Invisible Work: A Conversation on Craftsmanship, Curiosity, and Code

An interview with Konrad Szyc, Software Developer at Apsis

Most of the time, when software is working exactly as it should, nobody says a word. No alerts, no tickets, no calls — just silence. And for Konrad Szyc, Software Developer at Apsis, that silence is the sound of a job well done.

Konrad works on some of Apsis's most established products — maintaining, improving, and safeguarding the systems that thousands of customers rely on every day. It's work that rarely earns a standing ovation. But as Konrad explains, that's rather the point.

We sat down with him to talk about what it really means to maintain software, how AI is reshaping his day-to-day, and why a Polish improv theatre group might be one of the best things that ever happened to his career.

Konrad Szyc, Software Developer at Apsis

"It seemed like an interesting challenge"

How did you end up at Apsis?

Konrad: "I think it was the right fit at the right time. After a long project at my previous company, I'd built up confidence in my skills and knew I wanted something bigger — a company where English was the working language, where I could meet people from different backgrounds. Apsis had that international energy I was looking for.

But honestly, what really sealed it for me was something I came across while getting to know the company. On Efficy's own site there was a quote from a colleague about the no-blame culture here — the idea that when challenges arise, you look for solutions first, not someone to point a finger at. That resonated with me deeply; I think trust is the foundation of everything in a team.

And here's the part I didn't expect: I now actually work with the person who gave that quote. So I can say it wasn't just a nice statement on a careers page — it was true. I've experienced that culture firsthand, and there's a real sense that people are genuinely engaged in helping each other move forward. That's an uplifting thing to experience."  

The job that nobody notices, until it breaks

What does a typical week look like for you?

Konrad: "Honestly, there's no such thing as a typical week. Some weeks are about putting out fires. Others are about stepping back, planning ahead, grooming the roadmap in Jira. The surface area is quite wide and as a developer responsible for more than one product, there's a lot of ground to cover.

What surprises people most about my job is how much of it isn't actually coding. A large part of my time goes into organising tasks, estimating effort, creating runbooks, coordinating with the support and hosting teams. And when I am working with code, much of that time is spent reading it, not writing it. To understand what a long-standing piece of code does, you need patience and attention. You can't rush it."

What do you wish more people understood about software maintenance? 

Konrad: "Two things. First: most of the work is invisible when it's done right. If the software is running smoothly and customers never notice any disruption, that is the success. It's a strange kind of victory that you can only really celebrate internally.

Second: software is never finished. It's less like a product and more like a living thing — it needs to be tended to, adapted, kept healthy as the world around it changes. Those updates your phone prompts you to install? Someone is responsible for those. And for online services, they happen transparently, you never even see them. That's exactly the goal."

Embracing AI without losing the plot

How has AI changed the way you work day to day?

Konrad: "It's made me dramatically more effective. What AI does for me, more than anything, is take away the tedious parts of the work. Without it, you spend hours sifting through code and documentation, hunting for the piece of context you need. AI dives into all of that quickly and comes back with a summary you can then check. That's a real time-saver.

Where I've seen the biggest boost is incident response. When something goes wrong and you need to act fast, AI allows you to let it do the digging whilst you do the thinking. You're not distracted by trawling through logs or documentation. You can read the summaries and take decisions. And in incident response, speed of decision-making really matters, because this is the part customers will notice.

Beyond that, it helps with documentation, creating runbooks, getting a second opinion on ideas. You can throw a problem at it and see what comes back.

I think of working with AI a bit like driving a fast car that goes precisely where you point it — and you learn quickly to be careful where you point it. Then you find you don't always need to be at the wheel: you can set the destination and let it drive a stretch while you watch the road. The step beyond that is where you're not driving at all — you're directing several of these cars at once, each handling its own task.

That's the part I've been thinking hardest about lately: not just how I personally use these tools, but how a whole team — or an organisation — could be structured to work alongside them. I've built a working prototype that shows one way it could hold up technically — a proof the idea works, not the final word on it. With it, I've already had several AI workers running side by side on real tasks. And with that kind of leverage, human judgement matters more than ever, not less — because the hard part is no longer doing the work, it's deciding what good looks like and keeping everything pointed in the right direction. That part stays firmly human."

Where would you deliberately not use AI?

Konrad: "For me, there are a few firm boundaries. Anything directly modifying production infrastructure. Final approval on a code change before it ships. Deciding what work matters most and in what order — that one reflects values and context, which AI doesn't have. None of those are things I'd hand off, at least not now.

Beyond that, I think this question taps into the bigger anxieties around AI right now — the headlines about errors, deleted data, systems going wrong. I'd offer a more reassuring view, and not a purely theoretical one. My own stance is that these boundaries shouldn't rest on willpower or on trusting the tool to behave — as much as possible, they should be built into the setup. I've done a lot of that isolation work myself, and I'm a strong proponent of running the AI in a sealed-off space, kept away from the keys to anything truly critical, with the limits enforced by the system rather than by good intentions.

It's not absolute, though, and it shouldn't be. In a few places I've deliberately given it more room — where a mistake would be visible and easily undone. That's an accepted trade-off, not an oversight. The point isn't to lock everything down; it's to be clear-eyed about where the risk sits and what it would cost. Most of what people worry about with AI is risk our industry already knows how to contain. The harder limit lies somewhere else.

What AI genuinely cannot replace is human contact — being in the room with other people, present and responsive in a way no model can imitate. That's a big part of why stage improvisation matters to me."

The craft of staying curious

What keeps the work interesting after years in the industry?

Konrad: "The landscape never stops changing. Technologies arrive, others fall out of use, new approaches emerge. What I find genuinely exciting about maintaining established software is that you can take something that exists and apply new principles to it, practices that weren't available or simply weren't popular when the software was first written.

For instance, some of my work involves bringing modern monitoring into long-standing systems — better visibility into what the software is actually doing and why, so when something goes wrong you can pinpoint it fast. The tools for that have come a long way, and they've transformed how we understand these systems and respond to incidents. That kind of evolution keeps things fresh.

And right now, of course, the whole AI movement is turning everything on its head. The work has become more dynamic, more interesting — new challenges, new questions. That energises me."

What's underrated or underappreciated in the world of software right now?

Konrad: "The boring parts, and I mean that with respect. Security, infrastructure, software architecture. These take time, they're often invisible, and they don't generate headlines. But without them, nothing works properly. I think as the excitement around AI grows, it's worth remembering that the fundamentals haven't gone anywhere. If anything, they matter more."

Improv theatre and the soft skills developers don't talk about enough

You mentioned wanting to improve communication and soft skills earlier in your career. How have you worked on those?

Konrad: "I've been a member of an independent theatrical group for some years now, and I'm very much into stage improvisation. It started after I saw a TV show featuring an improv group and was inspired., I found a local open meeting through Facebook events and got involved from there.

What improv has given me is hard to replicate any other way. You learn to interact with people under pressure. You learn to accept that unexpected things will happen, because in improv, the unexpected is literally guaranteed. And you learn to be comfortable with that, rather than rattled by it.

There are also real parallels with the workplace. In improv, a good performer's job is actually to make their partner look good. You accept what they offer and build on it, rather than shutting it down — we call it 'yes, and'. That's what great teamwork looks like too.

Practically, it's also helped with things like speaking clearly, projecting confidence in your voice, managing nerves. You learn to accept a certain level of stage fright rather than fighting it, and that translates directly to situations like presenting in front of colleagues or leading a discussion in a difficult moment."

What advice would you give to a developer starting out today?

Konrad: "Learn to read code, not just write it. Especially now, with AI generating more and more code that developers then review, the ability to understand code by reading it is becoming one of the most important skills you can have.

Invest in software architecture —, understanding how to structure systems, how the pieces fit together meaningfully. I have a metaphor I like for this. Picture a large storage room. Throw everything into one big box shoved on a shelf, and finding anything becomes a nightmare. Now picture a modern set of drawers instead: everything in its labelled place, each one gliding open the moment you reach for it. And when your needs change, you can add whole new sets of drawers — or take away ones you no longer need — without disturbing the rest. That's what good architecture is: not just neat storage, but a system that adapts as you go. Get it right, and the software stays maintainable as it grows.

And then there are the soft skills — communication, trust-building, networking. I know there's a stereotype of the developer as a somewhat solitary figure, but in my experience, the relationships you build at work are fundamental to everything else. They show in your interviews. They shape your reputation. They follow you throughout your career."

The success that travels with you

What would you consider your biggest career success?

Konrad: "I think it's the level of trust I've managed to build with the people I work with. It grows throughout your career, and you carry it with you. Feedback from colleagues and from management tells me this is something I can genuinely be proud of.

When I came to Apsis, I arrived with real confidence — in my previous role I'd been trusted with important responsibilities and handled them well. That kind of personal track record — a personal brand, in marketing terms — is what I consider my biggest success. It doesn't belong to any one company. It's yours, and it travels with you." 

Interested to join Apsis and efficy Group? Check out our job openings.