Applied AI Engineer
It’s easy to build. It’s hard to make it actually work.
I build AI systems and products that have to work in the real world.
I’m drawn to unclear spaces where the technology is moving fast, the value is still forming, and the job is to turn possibility into a working system.
I’m especially interested in agentic systems, AI-native products, and systems that build systems.
Building AI tools, agents, and automations for use inside O2 Slovakia.
Working hands-on with LLMs, agentic workflows, APIs, integrations, sandboxing, DevOps, and Azure deployments — from experiments to running systems.
Helping build a shared AI environment where teams across the company can safely use AI, automate workflows, and develop their own solutions.
enterprise AI · current · o2.sk (opens in a new tab)
I build AI-native products and agentic systems, from the first idea to working software.
My work includes building agent harnesses, connecting models to tools and APIs, and experimenting with AI agents that can write code, run tests, deploy applications, and operate products.
Increasingly, cloud agents such as Hermes are taking on more of this work, turning JackoAI into an autonomous software factory.
I also help teams move beyond simply using AI tools toward building and working alongside AI systems.
independent practice · jackoai.com (opens in a new tab)
I built MyHoursLog because the software I was using for time tracking kept getting more complicated while the features I actually relied on were getting lost.
So I stripped the problem back to the essentials: the small set of functions I used regularly, plus the few things I always wished were there.
The product itself is intentionally simple. What’s more interesting is the experiment around it: using Hermes, a persistent AI agent, to explore how a small software product can be marketed, distributed, and improved with increasing autonomy.
I’m interested in what happens when one person can build a useful product and then give increasingly capable agents responsibility for helping it grow.
preparing for launch
What if one person could investigate problems that would normally require an entire research team?
RSI started with that question. I’m building a system that takes an ambitious objective and turns it into hypotheses, research plans, experiments, and evidence-driven decisions.
The system is designed to separate scientific reasoning from execution, preserve what has actually been validated, and progressively move promising ideas from analytical research toward real-world experimentation.
The core engine is working. I’m now turning it into a product that independent researchers, small teams, and business owners can use. I’m particularly excited about small and medium-sized manufacturers who could test promising ideas directly using their own production capabilities.
open-source release planned · product in development
It started with a book.
After finishing one of Karin Smirnoff’s novels, I wanted to read more stories like it. So I asked ChatGPT if it could write one for me. To my surprise, the result was incredible.
But there was a problem. AI could write compelling scenes and chapters, yet struggled to maintain coherence across an entire novel. Characters, events, and storylines would eventually lose continuity.
That was the moment I decided to build Shadow Writer.
What followed was over a year of evenings and weekends, working on the engineering and product experience needed to turn that initial experiment into a real consumer product.
Shadow Writer is now approaching its first public release. It’s my first major consumer AI product and an important step toward building an independent software business.
pre-release · myshadowwriter.com (opens in a new tab)
PrivaMind started with a frustration: constantly having to explain myself, my projects, and my context to different AI platforms.
But the bigger question was ownership. I wanted my ideas, intellectual property, memories, and personal AI infrastructure to remain under my control — ideally running locally, on my own hardware.
That doesn’t mean turning away from frontier AI. Quite the opposite. The idea is to combine private, locally operated intelligence with the best available models, delegating isolated tasks when needed without unnecessarily exposing the larger context.
Over time, this has evolved into a vision for an independent AI operating layer where memory, identity, permissions, and project state persist while models, agents, and interfaces come and go.
Surfaces change. State persists.
PrivaMind is a long-term, non-commercial, open-source project that is gradually taking shape as the technology catches up with the vision.
research & development · open-source project in progress
In October 2024, I decided to try something that had seemed impossible just a month earlier: building and launching an entire software product on my own, with AI as my engineering partner.
There were no autonomous coding agents doing the work. I started with v0 for the interface and worked directly with Claude through Anthropic’s console, manually moving code between conversations and my development environment.
Six weeks, roughly 350 hours, and 200 million tokens later, PickMyNiche was live — with authentication, payments, a production database, and a Python backend connecting multiple data sources and AI models.
The hardest part wasn’t generating code. It was making everything work together: integrations, security, error handling, testing, and deployment.
That experience fundamentally changed how I thought about building software. For the first time, I could turn my own ideas into working products without assembling an entire development team.
It also taught me something I’ve carried into every project since: AI makes building increasingly accessible, but turning software into a functioning business is an entirely different challenge.
launched October 2024 · pickmyniche.com (opens in a new tab)
The idea came to me while learning to sail.
Monohull sailboats have surprisingly little comfortable outdoor space, especially when anchored in a beautiful bay. I kept thinking how wonderful it would be to have somewhere to lie down beside the boat, relax above the water, or spend a night under the stars.
So I decided to build it.
SeaNest is a small family venture. ChatGPT is helping me design the product, while my mother will sew the first prototypes in my grandfather’s old workshop, which we’re turning into a small sewing studio.
Once we have a tested product, the plan is to expand production and use Hermes, a cloud AI agent, to help run the online store, marketing, and growth.
I’m excited about combining AI-assisted design, family craftsmanship, and autonomous business operations to bring a physical product to market.
And honestly, I can’t wait to see the first customers enjoying it on their boats.
physical product · prototyping · website & shop coming soon
I don’t think of AI simply as another tool or a replacement for human intelligence.
I’m interested in what happens when increasingly capable AI systems become integrated into how we think, create, learn, and make decisions.
Not humans versus machines. Something new emerging from the collaboration between them.
AI has changed the way I approach engineering. What once required manually writing and integrating every component is increasingly becoming a process of defining objectives, designing architectures, directing agents, and validating their work.
The interesting shift isn’t just faster development. It’s how much more one person or a small team can accomplish.
I’m experimenting with persistent cloud agents and custom harnesses that can support development, testing, deployment, operations, and growth.
The long-term vision is a software factory where people define direction and priorities while increasingly capable systems handle more of the execution, experimentation, and improvement.
JackoAI is my laboratory for exploring that possibility.
I believe Europe has the talent, research, engineering, and resources to build globally significant technology companies and infrastructure.
I want to contribute to ambitious projects that turn those capabilities into working systems and real-world impact.
Made in Europe
SVK · 2026