From jazz pianist to AI founder: A conversation with Mark Reinke

Interviewer: Olya Vvedenskaya

We sat down with Mark Oliver Reinke, co-founder and AI engineer at scienceOS, to talk about his drastic career turns, building meaningful technology, and what it’s really like to build a startup from the inside.

Olya: Mark, can you start by telling us who you are and what you do at scienceOS?

Mark: I would call myself an AI engineer or data scientist - one used to say data scientist, but it’s all centered around AI now. So I’d say an AI engineer.

At scienceOS, my main role is to develop and improve the AI agent. I work on prompts, on evaluating the agent, and on developing the tools that the agent uses. In our case, the agent uses tools to retrieve documents, publications, papers, and so on. That’s my main focus at scienceOS.

Olya: Your background is pretty unique, as far as I know. How did you get here?

Mark: My background is twofold. I used to be a musician - I studied jazz piano and practiced as a jazz pianist and music teacher until my mid-thirties. Then I decided to study computer science. I completely switched careers. In the first years, I worked in both fields in parallel - I taught at the Hochschule für Musik in Rostock, Germany, while studying computer science in Dresden, Germany.

After that, I started my first job in computer science. I still worked on one project as a musician, but it was hard to maintain both continuously. So for the last few years, I’ve concentrated only on tech.

Olya: What made you decide to leave music and make such a big change?

Mark: Well, obviously, it’s hard to be a musician or actually to be an artist of any kind. It’s hard to make a continuous living and to have a stable life. I noticed it was too tiring for me and that the bad sides were starting to outweigh the positive sides.

Although I have to admit, being a musician is an extremely rewarding profession. You tour, you give concerts, you make music - there are so many beautiful things. But it’s also really tough to always make your living, and to make a good living, you often have to do things you don’t really want to do, like teaching (in my case). I did a lot of teaching, and I didn’t really want to end up as a full-time teacher.

So I decided to do something safer, where things are a bit easier. At first, I was interested in psychology, but I realized it would take far too long before I could work in that field. So I went for something that was in demand and where I figured out I’d be good at it - and that interested me.

Olya: Let’s talk about how you met your co-founders. When did you meet André and Henri?

Mark: I met André pretty soon after I finished studying computer science. I started a job at a Dresden company that developed a search engine for companies, and that’s where I met André. After maybe a year, we started working closely together because André was responsible for all the smart things happening in the search engine. He was already working on machine learning back then, making the search engine more intelligent. That was very interesting to me because my focus was already on machine learning and data science.

Later, we got a big project for the German Patent Office and developed what we called “cognitive search” - a semantic search engine to identify whether patents were actually novel. We also built classifiers to determine which category a patent belonged to. These were really demanding and interesting projects, and we even published a paper about it.

Eventually, the idea came up to found our own company. We first founded Exdatis and worked on e-commerce, which was a bad idea. It didn’t work.  Then, when ChatGPT came up, we started giving AI workshops and building solutions with LLMs for companies. And through one of these projects, we met Henri. One day, Henri called and shared this idea, which was basically the idea of scienceOS. He was writing his MBA master’s thesis about it and asked if we wanted to join him. The rest is history [Mark is laughing].

Olya: What were your first impressions of André and Henri?

Mark: André was a very quiet guy, but I figured out very quickly that he was very smart and very creative. Since I was new in the company, I decided, “This is my man - I have to work with this guy.” And I knew interesting things would happen.

Henri also struck me as a very intelligent person with a very clear focus. André and I had worked together for many years, and we’re both people for whom everything is interesting: “Oh, that’s interesting, let’s do that.” Henri is much clearer about what he wants - he has a clear and very well-determined vision and goals. That was exactly what was missing. Henri brings clarity, marketing, sales, and interpersonal skills; he’s much more outgoing than André and me. I saw a lot of potential in that combination, and I think I was right.

Olya: When Henri pitched the scienceOS idea, how did it feel for you?

Mark: I was in immediately. There was not much hesitation. We were actually waiting for a project like this. As I mentioned before, we previously had worked on an e-commerce product, which wasn’t a good value and personality match for me at all. Turned out I have no interest in e-commerce. When Henri came with the proposition, it was immediately clear: this was the perfect match in terms of vision, mission, and values. This is what we want to do.

Olya: What made scienceOS feel like the right fit? Connection to science and research for you?

Mark: Not so much from the science side - I did study, and I even wrote a publication, but I never practiced serious science. For me, the match was more than that; it perfectly met my interests: search engines, retrieving knowledge, organizing knowledge, and finding knowledge.

This is kind of the perfect stage to realize all these things I’d always been interested in and worked on. And it was also the perfect time technologically. The idea came a couple of months before ChatGPT, and then ChatGPT came - and suddenly things that were very hard to realize before became much easier. Things that one never thought would be possible are possible now. It’s very rewarding that things I develop actually work now, and that they’re really helpful for people.

Olya: Was it all smooth sailing after that?

Mark: No, of course not. There are always ups and downs.

The first tough moment was when we realized that our original plan - getting investors and funding - wasn’t going to work easily. The market wasn’t good, and we’d have to put a lot of energy into fundraising, so we decided to do it ourselves using a bottom-up approach and sales rather than investment.

We had already invested two years in the e-commerce idea before. With scienceOS, it’s been much more successful, but still, it’s taken years of work without much financial return. That was tough.

But we could always see that people liked what we were doing and were willing to pay for it. There was always this feeling that there’s real potential - and that hopefully someday, we’ll be rewarded for all the hard work we’ve put in.

Olya: Has building scienceOS changed you as a person?

Mark: Everything one does passionately changes them as a person, of course. You learn a lot about how you work on problems, how you develop a product that’s in sync with what users need, and how you work together as a team.

We’ve learned to talk about issues but still get things done - because you can also get lost in talking about everything all the time. You have to find a balance between talking things through and getting things done. I think we’re pretty good at that.

Olya: Was there anything you had to unlearn along the way?

Mark: Definitely - especially in how software is built. There’s been a huge revolution in the last half year or so. We now use AI to build software. Since tools like Claude Code were released, how you write software has completely changed. You used to write code yourself. Now you very rarely write code - you tell the AI how to write the code. You’re not the coder anymore, you’re the manager. You have virtual programmers working for you.

That’s a massive shift, and it’s problematic in some ways - especially for junior developers. You still need to be the architect to review what the AI does, and you only get those skills if you’ve done this for some years. The AI still does a lot of rubbish - it needs experts to control it.

But for me personally, it’s amazing. Our interest isn’t in writing beautiful code. Our interest is in building a great product, and now we can do that much faster and more smoothly.

Olya: How did you feel when you first realized this shift?

Mark: Absolutely ecstatic. It’s just incredible. You’re no longer held up by stupid details that take hours or days - you just tell the AI to fix it, and it does it. For us, it’s amazing.

Olya: Is there something deeply personal about working on scienceOS for you?

Mark: Yes - it’s the feeling that you’re building something people actually use and that helps people. You get emails, feedback, people telling you they love it - that feels great.

And it is especially amazing because I believe what we’re doing is something positive.   We are creating something of worth. Even though I’m not a scientist, I believe in science, and I think it’s great that we can help scientists with something that’s genuinely positive.

Olya: If we asked your team, what would they say is your most annoying trait as a founder?

Mark: Probably that I’m very impatient. There’s a dynamic where André and Henri are more detail-loving people, and I’m someone who wants to take big steps right away. I want to improve the product in ways everybody notices, and I don’t want to focus on something small that is hidden in the last line of the update note.

That impatience can sometimes make me emotional or not as sympathetic as I could be. So yeah, that’s probably the most annoying thing.

Olya: Are there things you refuse to compromise on - even if it costs money?

Mark: Personally, yes. After being a founder for five years, it would be very difficult for me to work in a company again. I was a musician before, so I have this in me - I want to be my own boss.

I want flexibility, the possibility to realize my own ideas, and to have immediate feedback. You create something, and people use it and tell you what they think. In a large company, that can be very indirect - you don’t always know what the meaning of your work is. Here, there’s a very direct link between what you do and the effect it has.

Olya: Even when things are hard, is it still worth it?

Mark: Oh, definitely. Absolutely. There are ups and downs, but in general, I love what I’m doing. It means a lot to me. It’s very rewarding. I also do need that feeling of growth, the feeling that things are getting better. There was a time last summer when things were sort of quiet because people were on holiday and universities weren’t really active in using the tool. That was tough because I started wondering, “What if it stays quiet forever?” It never really went bad, but this quite flat line in usage was already scary. 

So I do need that feeling of upward trajectory when I am working hard on something.

Olya: Any advice you’d like to share?

Mark: I would definitely encourage people to follow their dreams, to found companies, and to work in startups - especially nowadays. There’s no real security in a lot of jobs anymore, and at the same time, it’s become so much easier with AI to build things and create companies with very few resources. You can do it even alone; you can run a company with AI.

That said, I started a startup journey in my mid-40s, and I wouldn’t recommend that to everybody. It’s tougher than I thought. It’s hard. If you are in your mid-30s, I’d say: go for it. In your mid-40s, it’s a tougher decision - but for me, it was worth it.