Head of Software Development Department at Innowise Ivan Shatukha – on how not to fear AI, not to devalue people, and build a system that scales faster than the market.
The artificial intelligence market is experiencing not a boom, but a shakeup. While some companies are buying up models and launching hundreds of demo projects, major players like OpenAI and Meta publicly admit: expectations are too inflated, and most generative AI projects don’t generate revenue. By various estimates, up to 95% of such initiatives are unprofitable.
Against this backdrop, the experience of Ivan Shatukha is particularly interesting – co-founder and Head of Software Development Dераrtmепt at IT company Innowise, where AI is not hype, but a working tool. Ivan didn’t just build AI products. He created a system that allows scaling teams and directions, from Python to Big Data and machine learning, times faster than the market. Without promises of magic. Only through processes, architecture, and common sense. In this interview, he’ll explain why he’s not afraid of AI, how to scale not just code but people, and why the market has already begun dividing into those who build systems – and those who make presentations.
– Ivan, today, many experts are saying that the AI revolution has been overhyped, and that many generative AI projects have caused concern and not always brought profit. Do you see this as a crisis or as a natural part of the process?
– This is not a crisis, but rather a maturing market. If you look at the development of digital startups in the early 2000s, there was the same dynamic. First euphoria, followed by crash, and only after that the emergence of sustainable companies that know not only how to promise, but how to deliver. Right now in AI we’re going through the same funnel. And personally for me this is good news. Because those who built products for the sake of presentations get filtered out, while those who build working systems remain. Most people’s problems arise not from technology, but from not understanding what they need it for. When you make AI an end in itself you lose. When you use it to solve a specific problem you gain an advantage. For me AI isn’t a separate entity. It’s just another level of automation that can be integrated into processes as naturally as CRM, ERP or BI were once integrated. If you know how to build business processes you’ll figure out how to integrate AI into them.
– Even before taking your current leadership position at Innowise, a large IT company, you had already been involved in the implementation of AI there – for example, starting as a developer you created an AI-based tool for doctors in the VOKA project, within which you implemented your revolutionary AEM-GDD Model of architectural expertise management. Is it possible to say that AI has already proved its effectiveness to you, and you have no doubt about the work of AI tools?
– Yes, absolutely. When you see once how technology works not on a demo, but on real pain you no longer have doubts. One of the key examples is the VOKA project, 3D anatomy & pathology atlas, that we started as an experiment and developed into a full-fledged international product. It helps doctors from students to surgeons visualize anatomy and pathologies in 3D, model complex operations, and prepare for procedures that previously required hours of work with textbooks, images and colleagues. In this project I created and implemented my own architectural model – AEM-GDD (Architectural Expertise Management – Guided Design & Development). It is a three-level framework that captures and structures architectural expertise, turns it into a living knowledge base, and then guides both design and development. The key idea is “authorial expertise”: every decision, pattern or rule is tied to a concrete expert, not just to “the system”. Thanks to this approach we achieved measurable results: time-to-model dropped by 70%, release frequency doubled, and onboarding time for new engineers was cut by more than half. Now VOKA is used in 190 countries. But for me its value isn’t in coverage, but in the principle itself. This is a story about how AI doesn’t displace a specialist, but amplifies them. It’s not a doctor’s replacement. It’s a tool that saves time, reduces risks and helps make decisions more accurately. And this is exactly where I see the real value of AI; it doesn’t take away a profession, it makes it stronger. VOKA was a challenge for us not commercial, but engineering. We wanted to prove and we did.
– In your opinion, AI is not eliminating jobs, but rather transforming them. How has this implementation helped you expand the Innowise team from three people to hundreds?
– Absolutely. Look, I honestly don’t understand where this fear comes from. AI custom software development doesn’t replace people. It replaces slow people. That’s a huge difference. We’re not building neural networks that write code instead of humans. We’re building architecture where smart engineers can focus on what matters, instead of burning hours on stuff that should be automated. We’re not getting rid of people. We’re getting rid of all the noise around them. Today we have 2,500 employees in the company, but we also maintain an active pool of 70,000 candidates we’ve already sorted by tech stack, experience, region, and language. This isn’t just some resume database. It’s a system where when a client needs something like five Python developers in Central Europe, we can deliver a shortlist in three days. Not three weeks. Not “we’ll start looking around.” We deliver. And that’s not marketing talk, that’s actual engineering at work. Not because AI replaced our recruiters. But because it stripped out all the manual busywork: matching tech stacks, checking skill levels, tracking when people are available. The recruiter still makes the call. But everything that could be automated? It’s automated.
– Besides AI, as you have already mentioned, you have implemented effective business processes, automated workflows and introduced new technological directions such as Python and Big Data. Why exactly did you focus on them, and what did it give the company?
– There was nothing random about these decisions. We chose Python and Big Data because they’re not just technologies, they’re the infrastructure languages and approaches that modern automation runs on. Python gives you flexibility; you can build internal services and ML models with it. Big Data is the foundation of any serious AI work. Without data, any talk about models is just theoretical physics. When you go deep into something instead of just “this is trendy, let’s do it,” you build a platform you can lean on for other projects. That’s exactly what happened. We started with individual pilots, then turned them into full-blown divisions, and then they started bringing in clients. Not because we were shouting about having “AI inside.” But because we could actually solve real problems: business analytics, process optimization, training models for specific use cases. If we’d gotten stuck with standard web tech and never moved into Python and big data, I think we would’ve just stayed another cookie-cutter company.
– Your approach is truly unique, as it is copied and implemented by your competitors. Is this a challenge or a recognition for you?
– It’s a sign we’re in the right direction. If people are copying you, you’re ahead. The question is how fast you can keep moving further out front. When you solve a genuinely tough problem like the shortage of skilled specialists in the market and build an approach that actually works, it makes sense that others start borrowing those mechanics. But copying the technology is one thing. Copying the culture, the discipline, the rhythm that’s completely different. We built everything as a system: from recruiting to team growth, from internal training to external scaling. You can’t just reproduce that from a playbook.
– Your experience is being adopted by both colleagues and competitors, and this is also part of your recognition. You share your methods through multiple channels like public speaking at industry events such as OpenIT, Start-IT, and GeekStorm. How important is mentoring and experience transfer for you?
– It’s a strategic part. We’re not building a company where all the knowledge sits in two people’s heads. We’re creating an environment where expertise spreads horizontally. Conference speaking, mentoring, internal academies, public content — this isn’t about PR. It’s about accelerating company growth because people are constantly running into new practices, external challenges, real problems. Speaking at OpenIT, which has been held for more than 10 years and has grown into one of the largest national IT conferences with several thousand participants annually, is both prestigious and highly responsible. The event gathers leading experts, entrepreneurs, and developers from across the country, making it a key platform for exchanging real cases and working practices. For me, it’s not about being on stage, it’s about the chance to share knowledge with a broad professional community, which then multiplies its effect inside companies and across the market. I also mentor startups and judge competitions a lot, not because I have tons of free time, but because in those formats you quickly see which approaches work and which don’t. Then you can bring that back into the main business. This isn’t about “showing off expertise.” It’s a way to stay on the cutting edge – not chasing the market, but shaping it.
– But surely it is equally important to train the team. How do you implement this?
– It’s a requirement for scaling. Without a learning environment, you either lose quality or lose speed. We approached this systematically. We built internal infrastructure where an employee goes from intern to full engineer on real projects. This isn’t simulation or “playing around with code”, this is real challenges, mentoring, internal certifications. We produce internal courses, create educational content we’re not embarrassed to show clients. We have certified Python and Big Data teams, and that’s not just some formal piece of paper, it’s proven capability that lets us take on serious projects. When you show a client you have systematic training inside your company, they have fewer doubts about whether you can keep up with the pace they’re used to.
– In your training, you prepare specialists for future changes and teach them to adapt to the demands of AI. What do you see as the future of AI in general, and what should businesses and specialists prepare for?
– I think AI will stop being “the thing”. It’ll just become background. Like the internet did at some point. It’ll be built into every process: from medicine to education, from sales to logistics management. But the winners won’t be those with the trendiest neural network, they’ll be the ones who’ve built the ability inside their company to learn fast, adapt fast, scale fast. To those who are scared today, I’ll say this: AI won’t replace your job, but it will change the way your role works. The key is not resisting the shift, but learning how to move with it. Companies that build the ability to adapt will thrive; those that wait for things to go back to “normal” will fall behind. AI won’t replace you, but someone who knows how to use AI effectively might replace your whole business.
