From Research via a Startup to Škoda Auto. "The Golden Age of IT Is Not Ending, but It Is Changing Its Standards," Says FIT BUT Graduate

He only graduated from FIT in 2024, yet he already has a wealth of professional experience – from academic research and a leadership position in a startup to working for a multinational corporation. Today, his work primarily revolves around Artificial Intelligence. "It’s an extremely dynamic field, but you can't succeed in it long-term without solid foundations," says Petr Pouč, who works as an AI Engineer at Green:Code (Škoda Auto).

You graduated from FIT only recently, yet your CV already includes research, a startup, and a corporate role. How did you manage that?

My career didn't start with my diploma. It’s good to find practical experience while still in school. At FIT, there are many research projects, and numerous IT companies collaborate with the faculty to offer internships. I took advantage of both opportunities. After my final exams, I didn't start as a complete junior; I didn't have to start from scratch and could immediately negotiate more interesting terms.

During your studies, you joined the FETA research group. Was this the moment that steered you toward AI?

Definitely. That’s where I first truly delved into the depths of machine learning, which I now see as a key component and the foundation of AI. I originally wanted to gain some initial work experience, but it ended up showing me the direction I wanted to take for my entire career. Research allowed me to connect theory from lectures with real-world data, which I later utilized in my master’s thesis. I also co-authored an article in the renowned journal Data in Brief. All of this helped me significantly when entering the job market – having a project for the Ministry of the Interior on your CV as a student is a great conversation starter at any interview.

Parallel to your research at FETA, you joined the Brno-based startup Lakmoos, which uses AI to create "digital clones" of customers for market research. And you started straight in a leadership position…

That was a baptism by fire. I joined as a Data Science Team Lead with minimal work experience—and even less in people management—and suddenly, I had to lead a team. This was where I first felt the real benefit of FIT. Even though everything was new to me, the ability to learn quickly, react to changes, meet deadlines, work in a group, or lead a project… these are all skills you hone daily at FIT. The beginning was quite bumpy. Requirements changed, team members rotated, and we often worked overtime and weekends. But eventually, everything stabilized, the company survived, and today it’s doing very well. I took away the ability to lead people, communicate with customers, and manage my time effectively. I think every developer should try a startup at least once; compared to a corporate environment, it’s a completely incomparable experience.

You are currently working for a corporation—you moved into the structures of Škoda Auto, specifically the joint venture Green:Code, where you work as an AI Engineer. What does such a job involve?

The work of an AI/ML Engineer varies greatly depending on the environment. The transition from a startup to a corporation meant a significant change for me, especially in terms of dynamics and pace. In a startup environment, there is an extreme emphasis on delivery speed. If the deadline is Friday, we want it done by Wednesday. In contrast, the corporate environment is characterized by much longer approval processes, and deadlines are viewed with a larger margin. A project with an original deadline in October might easily move to December, and ultimately, completion in March is considered a success.

Regarding the actual workload, my role has evolved over time. Previously, in research, I focused on neural network architectures—going deep and examining how models behave. Today, in a production environment, we operate at several levels of abstraction higher. We don’t primarily develop models from scratch; instead, we orchestrate powerful pre-trained models, such as those from OpenAI or open-source variants like Llama. The role of a modern ML Engineer is much more about building a robust ecosystem around these models and their effective integration into company processes.

That sounds as if a programmer's job is becoming simpler. Does it mean you no longer need to understand what’s happening "under the hood" and just need to task the AI correctly?

On the contrary, the role is incredibly complex today. You must understand the entire pipeline—from document extraction and structuring to prompt engineering and managing vector databases like Qdrant, which involves working with embedding models, similarity metrics, and indexing for RAG architectures. You design scalable APIs that must handle thousands of queries and address rate limiting or fallback mechanisms. You must think critically—address why a model is hallucinating, how to systematically remove bias, and whether the priority for a given task is speed or accuracy. Is it exploratory analysis or production deployment? These decisions then affect the entire architecture of the solution. And what is perhaps hardest in this field is the absolute necessity to stay constantly updated. New technologies, libraries, models, and best practices come out all the time. Sometimes I feel that if I go on vacation for a week, I return to a world that has already moved on. It’s an extremely dynamic environment, which is the most exciting part of the job.

So it’s highly specialized work. Yet, headlines often appear in the media saying the "golden age of IT is ending" and that AI will soon replace programmers. What’s your take on that?

It’s true that new phenomena like "vibe-coding" are emerging, where people have code generated without deeply understanding it. However, this approach has clear limits when it comes to resolving errors, debugging unexpected behavior, assessing security risks, optimizing performance under load, or ensuring stable, scalable operation in production. I see it as the golden age of IT not ending, but the requirements shifting. A junior programmer who just mechanically writes code according to specs will have a hard time and might not be needed. Conversely, experts who understand systems in their full complexity—who can design robust solutions, identify risks, and ensure security and scalability—will still be in high demand. Perhaps even more than ever before.

So, FIT BUT and its legendary difficulty are still relevant?

Absolutely. FIT gave me a solid foundation – when you understand algorithms, data structures, network protocols, and the basic principles of software engineering, you can adapt to anything. Technologies change every year, but these principles remain. Especially in areas like cybersecurity, system architecture, cloud, networking, or embedded systems, the depth of knowledge really makes a difference. These are all areas where AI can help with individual tasks, but it will never replace a human who understands the full context and can make critical decisions.

You yourself are creating solutions for Škoda to make people's work easier. But where do you see the line between "assistance" and the moment AI completely replaces a human?

I hold the view that AI will not replace people. But people who use AI effectively will replace those who resist it. The department I work in focuses on developing internal AI applications to save people a lot of worry and time. For example, we are currently developing a "semantic checklist" for manufacturing. It goes through complex documentation for the technologist—checking everything from logic and grammar to formal requirements, the correctness of links, and compliance with manuals—and generates suggestions for specific edits. Or we are working on an internal RAG-based search engine that can find an exact answer to a technical question from tens of gigabytes of internal documents and highlight relevant passages. These aren't tools to take someone's job; they just make the work easier.

You say technology is changing before your eyes. Do you have a long-term career plan in such an environment?

Currently, I’m happy with the projects I’m working on; they have a real impact and keep pushing me forward technically. At the same time, I realize that the industry I move in is extremely dynamic—the boundaries of what is technically feasible are constantly moving, and entirely new areas are emerging. So, I don't have specific long-term plans; I feel they lack purpose. I believe that in IT, education is an endless process. It’s important not to stagnate, to continuously maintain an overview, experiment with new technologies, follow research, and be able to quickly adapt your solutions to what works best.

Was it a jump into the deep end for you, coming from school into such an environment? What was your transition from studies to work like?

To be honest, it was actually very pleasant. Intense preparation for final exams and packed exam periods taught me to manage my time effectively. After finishing my studies, for the first few months, I felt like my day suddenly had 30 hours (laughs). But today, I see that very difficulty and practical focus as the greatest benefit of FIT. The vast majority of subjects include practical projects as part of the credit; thus, all students have the opportunity to transfer their knowledge into practice. FIT also brought me many contacts, acquaintances, and friends, which is very useful in the IT field because these relationships often bring interesting job opportunities.

Let’s go back to the very beginning. Were you drawn to technology as a child, or did the idea of studying IT arise just before graduation?

I’ve been drawn to technology since I was a kid. Computers fascinated me. I remember my parents having to turn off the Wi-Fi or lock the bedroom with the router to force me to focus on school too. But the final decision was made in high school when choosing seminars. At that time, I couldn't yet imagine exactly what it entailed, but programming just sounded "cool" (hustě). Plus, it was taught by my favorite homeroom teacher, which played a big role.

Why did FIT "win" for you back then?

Back then, students from IT faculties gave presentations at the seminar and said: "If you want a quiet life, go somewhere else. FIT is a grind." I turned to my best friend and said how embarrassing it would be to choose a school just because it’s easy. I was attracted by the challenge. FIT had a reputation as a prestigious and demanding faculty, and I liked that.

Do you remember your first impressions of the faculty?

I remember it exactly. The first few days were intense; we were overwhelmed with information. What stuck in my mind most was a sentence from one of the instructors during the introductory lecture: "Look to your left, look to your right. Out of the three of you, only one will remain at the end of the studies." This "scaremongering" was important—it didn't allow us to slack off at the beginning. In the end, though, he wasn't quite right. My friends from both sides and I successfully finished our studies, even the Master’s degree (laughs).

---

Ing. Petr Pouč is a graduate of the Faculty of Information Technology at BUT in Brno (2024). He started his IT career during his studies in the FETA research group, where he participated in projects using machine learning. He gained management experience as a Data Science Team Lead at the Brno startup Lakmoos, where he still operates. Concurrently, he works as a Machine Learning Engineer at Green:Code (Škoda Auto), specializing in the implementation of AI tools into company processes.