— Developers are going through a tough period. After years of rapid growth and high demand, the industry is now facing layoffs, hiring freezes, and reduced benefits. How do you view this uncertainty, and what do you see ahead for IT professionals?
IT is a young industry, and we didn't have enough time to define our profession. At IT conferences, you can still hear a lot of discussions on what it means to be a software engineer and what kind of skills one needs for it. We are not yet adults as an industry. But now the big change is coming, the change that is caused by us, so we find ourselves in a moment where we need to grow up and redefine what we are going to do
A year ago, when ChatGPT and all the vibe coding hype started, I was in panic mode. I thought we were done. But then I started using AI coding assistance every day and realized it was just another tool
Explore all current job openings in Poland!
— But on LinkedIn, you wrote a post saying that agentic coding makes you dumb, lazy, and jobless.
This is the risk, and I'm scared that we are ignoring it. The whole idea of engineering is solving problems. To solve problems, you need tools. That's one thing. The other thing is that you need understanding. If you stop using your problem-solving skills and just type in prompts, you become dumb
From an evolutionary perspective, our brains are designed to optimize and to be lazy to save energy. But learning is a feedback cycle. Sometimes the feedback we get is positive; other times, it's negative. You learn programming by making tons of mistakes. The first five years of programming are one big sequence of errors. And you just must enjoy it. But with AI coding for us, we don't have this positive-negative feedback loop. When the AI proposes a solution to us, we cannot assess whether it's good or not. We'll just blindly accept it. In vibe coding, you do nothing, and you get a reward
This is the worst time for juniors. As a senior, because of my experience, I know exactly what I want from the system, and when I write a prompt, I can be really specific. I already know how to avoid all the usual pitfalls and create boundaries for AI. AI makes senior developers more productive but slows juniors down
— On the one hand, you want to use the best tools. On the other hand, you say the best tools make you dumb and destroy your problem-solving skills. So, what is the solution?
I try to limit myself. I'm selective about when I use AI and when I don't. This is kind of a diet. For example, if a project uses a technology I don’t know well because I rarely work with it, I use AI. Learning it from scratch would take a lot of time; I won’t need it again anytime soon, and I’d likely forget it within a week anyway. Another example is analyzing an old code I haven't seen for a year. It would take me a week to understand what's going on, but AI can explain and summarize it to me immediately so I can jump in quickly. I also use AI to do research and explore a topic. But I don't use it for coding. I find pleasure in coding. So why should I give my most interesting and fun part of my work to some artificial intelligence?
I give AI boring tasks and leave the fun part to me.
— So, you’re no longer worried that AI will replace you?
What can happen if AI learns from AI-generated code? Will it become better? I doubt it. There is a risk of a slow but steady decline in quality. So, we need a human in the loop
So, the problem for companies is how to train people if AI creates all the code, and how to train AI if no people are writing code
People are the major cost in every software development project. So, on the hype, companies have decided that they will not need people: fewer people, fewer problems, more money. But actually, you can have the same amount of work done with fewer people, or more work with the same number of people. What will companies choose in the end? The essence of capitalism is doing more. So, I think companies will probably realize soon that it makes more sense to combine the same people with AI
I also think what we are now seeing in the IT market is mostly a post-COVID world. During COVID, everybody was hiring like crazy. And we're still recovering and trying to balance things out after COVID. But of course, it's easier to blame AI
Until now, thanks to AI, we have become better at generating more shitty code. Now, the trick we need to learn is to be better at defining what we want
In the near future, programmers will be more like translators and validators, turning requirements into prompts. We will need to educate people so they have the capabilities and skills to write proper requirements and verify what was produced quickly. We also need to teach juniors to be architects who can think ahead about scalability, security, etc.
— Speaking about education, you have a higher scientific education, and you conduct programming classes. Do you believe universities are still important for programmers today?
The way the networks and the CPU work didn't change in all these years. In detail, we made huge progress, but the foundations of IT remain the same. And what I found is that understanding these foundations is super useful. Even writing your own programming language is the best exercise ever to understand the foundations. Whether it's Node.js or JVM, Rust or PHP, we are dealing with the same hardware, the same network problems, and the same laws of physics.
That is why I believe studying can be very useful, and I’d say it's better to study physics than math if you are a programmer, because we have far more problems arising from the physical world, like the speed of light. People are forgetting that the CPU is not some magical crystal but a piece of engineering, and that CPUs have limitations and bugs. You cannot ignore hardware. There is a piece of silicone that performs a specific function. It has its limitations. And if you don't respect what the CPU is doing, your program will be slow
You have to understand what's going on under the hood. If you have the foundations, you can assess what AI is generating. And you can be better at defining what exactly you want from AI. Otherwise, it's all dark magic to you
With AI, we no longer need to focus on specific language or specific disciplines. I can be full-stack, DevOps, or QA, just give me AI and enough tokens. But I need to really understand the logic of it. Universities and anyone responsible for education will have to start producing more people with general knowledge who can understand the process end-to-end, not necessarily with a high level of detail
— Why did you personally start programming in the first place?
My first computer was an Atari 8-bit with 64 kilobytes of RAM, and it was probably the only computer within a 50-kilometer radius
I sucked at playing computer games, while my younger brother was really good at it. So, I started learning a basic programming language so I could write a game that my brother couldn't beat me in. This is how I got into computers
But I never thought about making a career in programming; I treated it as a hobby. I studied chemistry at Krakow Technology University. When I graduated, it was a tough time in Poland, the end of communism, and I couldn't find a job as a chemist. Meanwhile, companies started looking for programmers. So, I decided to find something in programming for money, while I continued searching for a “proper” job, the job I was studying for five years
About two years later, I said, "Okay, that's actually fun to do," and decided to focus on programming and computers. I pursued postgraduate studies and applied for a programming position
The interviewer was the company's CTO, and we had an excellent discussion about programming at a higher level of abstraction, like architecture and design. After 40 minutes, he said, OK, I'm running out of time, but do you know Java? And I don't know what happened in my head. I said, Yes. And he said, Okay, you're hired. Fortunately, I had a 3-month notice period with the previous company, which I used to learn Java really quickly. And this is where things got serious
— How did you choose to specialize in Java virtual machines?
In a professional journey, having skills is one thing, but the ability to take risks is quite another. I am the risky one.
For a long time, I worked on SaaS platforms. And every time we were in situations where disasters occurred, like application outages, system crashes, and poor performance, everyone would hide; nobody wanted to fix them. While I would say, Okay, give it to me. I thought that's where I can learn. This way, I started slowly diving into application performance and benchmarking
But nobody really cared about performance until the cloud providers came. Nowadays, you can scale things by adding more hardware: if something is slow, you just go to AWS and spin up another bunch of machines. The problem is solved, and everybody is happy until the bill comes. So, what I'm mostly doing is finding ways to spend less money on AWS by writing faster software. There are hundreds of switches and configuration options that can speed up your code
The IT system is like one big onion. Everything interesting in IT happened in the seventies and eighties, and since then, for decades, we've built layer upon layer, and we haven't removed anything because we were too scared to break things. So, to fix performance and stability issues, we have to go deeper and deeper, sometimes down to the hardware level, even reading papers from the seventies, and that's when I feel alive: when I optimize the systems and make them faster. This is the job that makes me happy. I tried so many things, and now I don't see myself anywhere else
— What role does Poland play in the international IT market today, and how do you think this position will evolve in the coming years?
Maybe I should not complain, but complaining is part of being Polish. It's like in our constitution, you have to complain.
For a long time, Poland was like a programmer's farm. We were just a service delivery country. The product ideas were not created here. We were mostly developing software for something invented elsewhere. I worked for a couple of years for a US company, and it was really annoying and frustrating to have the architects and all the technical decisions made in the United States, while we were just delivering stuff
The coders and the scientists behind OpenAI and LLM are from Poland. These guys could have done it in Poland! This is making me angry
We have, in my opinion, the best IT community ever. If you go to Polish conferences, it's all crazy people talking about programming over beer. And even at various after-parties, when everybody's dancing, we are still talking about programming. People really love what they do
So, I think that's a great time for Polish programmers. We are in a good position, in great shape; we just need to figure out a way to be investors and entrepreneurs, driving more products from Poland
