AI that survives contact
with the real world.
|
I help companies put AI and custom software to work — from a small tool that kills a manual routine, to production ML systems running on a factory floor. 9+ years of building the unglamorous parts that actually have to work.
What I can help with
Most people find me somewhere between “we have an idea but no one to build it” and “this is broken and we’re out of ideas”. Both are fine starting points. Here is what working together buys you, and what it costs to find out.
Strategy: working out what's actually worth building
The hardest part usually isn't the code — it's naming the real problem. We map how the work flows through your business today, find where time and money leak out, and decide what technology (if any) is worth pointing at it. You get a plan you could hand to anyone, not just me.
- Sessions to map your current process end to end
- Where AI genuinely helps — and where it's hype
- A prioritised roadmap: quick wins first, big bets later
- Build vs. buy, and which tools are worth paying for
- Honest cost, effort and payback estimates
- A written plan you own, with or without me
AI in your business
You know AI could help somewhere, but not where to start. I find the one workflow worth automating first, then build it.
- Assistants that answer from your own documents
- Agents that handle repetitive back-office work
- Email, invoice & report processing
- Honest advice on what AI won't fix
Custom tools & automation
That spreadsheet everyone is scared to touch, the report someone rebuilds by hand every Monday — those become software.
- Internal dashboards & admin tools
- Scripts that replace manual routine
- Integrations between systems that don't talk
- Small web apps, built and handed over
Data & ML engineering
The heavier end: pipelines that don't break at 3am and models that survive contact with production.
- Forecasting, regression and predictive models
- Data pipelines & warehouse setup
- Deployment on Docker & Kubernetes
- Image and video measurement tasks
Consulting & second opinions
Sometimes you don't need a build — you need someone who has shipped this before to tell you if the plan holds.
- Architecture & code reviews
- Technical due diligence
- Getting a stalled project moving
- Team mentoring on data & ML
How working together looks
three steps, no surprisesA free 30-minute call
Tell me the problem in plain language. No prep, no slides. If I'm not the right person, I'll say so and point you elsewhere.
Fixed scope and price
Within a few days you get a short written proposal: what gets built, what it costs, when it lands. No open-ended hourly billing.
Built, delivered, yours
Working software with the source, documentation, and a handover session. You own it — no lock-in to me or anyone else.
Not sure which of these you need? That’s exactly what the first call is for.
Book a free 30-min call →Things I’ve built
Proof rather than promises. Paid engagements first — the systems running inside real companies — then the products I build on my own time, which is where you can actually click around and judge the craft yourself.
Client engagements
production systems, under NDA where it mattersTwo production systems on a working plant
A gas-blending optimizer for coke oven underfiring — pipelines unifying multiple plant databases and sensor feeds, regression models over high-frequency data, and a Django operator dashboard — holding heat targets while targeting GWh-scale annual energy savings. Alongside it, a real-time video pipeline whose UNet segmentation model sizes ore and dust on conveyor belts, replacing periodic manual sampling and flagging off-spec material as it moves.
Where to put an EV charger, across three continents
Built and owned a location analytics platform helping charge point operators identify the top 10% of candidate sites across Europe, the US, and the Middle East & Africa. The core model predicts kWh throughput at a candidate site from AADT traffic counts and local demand signals; around it sit geospatial ingestion pipelines pulling millions of records from third-party APIs, partner databases and custom scrapers, and Kepler.gl maps and Power BI dashboards used for investment decisions.
Behaviour systems for humanoid robots
Interactive behaviour for humanoid robots used in marketing and data collection: autonomous charging and movement, speech recognition and synthesis, and Google Sheets / SMS / email data sync.
Backend for the ECB's AnaCredit
Led backend engineering for AnaCredit, the ECB's granular credit and credit-risk reporting system — secure ingestion and validation modules for regulated bank submissions, an internal admin UI and REST API, with monitoring, logging and retry strategies for high availability.
Products & side projects
built on my own time — all live, click any of them
Saturn Lab
Self-hosted ML pipeline management — visual canvas, experiment tracking, and notebook orchestration for data teams.
PollenPal
Real-time pollen forecasts and personalized allergy alerts to help users survive pollen season.
SpotPilot
Live electricity spot price tracker for central European markets — hourly charts, price alerts, and consumption cost estimator.
MealBox
Box-tracking platform for canteens and catering companies — QR scan in/out, live dashboard, overdue alerts, and multi-location management.
FIRE Calculator
Financial independence forecasting based on income, savings rate, and target expenses.
Rent or Sell
Financial decision engine comparing property rent vs. sell returns and opportunity costs.
Portfolio Advisor
Asset allocation analyzer using Modern Portfolio Theory and risk-return optimization.
Politician Trade Tracker
Searchable database of 22,000+ stock trades disclosed by U.S. House members — red-flag detection, performance vs. the S&P 500, and directional signals when several members pile into the same ticker.
Black-Scholes
Interactive option pricing via Black-Scholes formula with live parameter visualization.
Netflix Calculator
Determines subscription value by actual viewing time and entertainment cost per hour.
LexiBoost
Fun language-learning tool for English and Italian with spaced repetition.
Chat Control Check
Register of chat and email services showing which ones the EU can legally scan under Chat Control — searchable, in four languages.
Pigeon Empire
Idle clicker game — recruit a flock, feed the pigeons, and build your empire one poop at a time.
The person you’d be hiring
The CV part, for anyone who wants it: where I’ve worked, what I actually know how to use, and what I do when I’m not at a keyboard.
Currently a Machine Learning Engineer at U.S. Steel Košice, building process-optimization and computer vision systems for heavy industry — a gas-blending optimizer for coke oven underfiring, and a real-time segmentation model that sizes ore and dust on conveyor belts. I own both end to end, from annotation and training to Kubernetes deployment.
Before that, six years as a Senior Full-Stack Data Engineer at Zoniq, building a geospatial analytics platform for EV charging networks — the model predicting kWh throughput at candidate sites, the pipelines feeding it millions of records, and the maps and dashboards clients made investment decisions on. Earlier still, regulatory backend systems for the ECB’s AnaCredit at the National Bank of Slovakia.
MSc in Intelligent Software Systems from STU Bratislava (thesis: community detection in networks), after a BSc in Computer Science at the same faculty. Based in Prešov. Outside work: weightlifting, running, books, and building finance tools that make data accessible to everyone.
I take on a small number of freelance projects alongside that — which means I’m selective, but the ones I take get proper attention.
Technical arsenal
what I reach for, roughly in order of how oftenGot something that needs building?
Book a free 30-minute call and describe the problem in your own words. You’ll leave with a straight answer on whether it’s worth doing, roughly what it takes, and whether I’m the right person for it. No pitch deck, no obligation.
Remote worldwide · English & Slovak · Usually reply within a day