I am Roland Nagy, a senior Shopify and full-stack developer. CheckoutLabs is my focused development practice for agencies, technical teams, and merchants that need senior Shopify capacity without hiring full-time.
I have been drawn to software since childhood, from early arcade games and a Commodore 64 to teaching myself programming at 16 from a Turbo Pascal book. The path was not perfectly linear, and software engineering was not an obvious route in the environment I grew up in, but the curiosity stayed.
That early interest eventually became a professional path through computer science studies, full-stack engineering, and a long-running Shopify specialization. Today I use CheckoutLabs to offer focused senior development capacity for teams that need someone who can handle the parts that do not fit neatly into a theme setting or off-the-shelf app.
Computer science foundations, ongoing AI studies, and practical software engineering experience behind the Shopify work.
Computer Science background at Eotvos Lorand University, with foundations in algorithms, databases, software engineering, web development, and applied programming.
Ongoing MSc studies in Artificial Intelligence, extending the software engineering background with machine learning, data-oriented thinking, and AI system design.
Years of Shopify delivery across custom storefronts, integrations, checkout extensions, configurators, B2B commerce, and ongoing production support.
Selected coursework and ongoing research from the MSc in Artificial Intelligence, kept separate from commercial Shopify work.
Built a multimodal (text + image) LLM that generates structured descriptions from images to a defined technical specification, implemented end-to-end in a Colab notebook.
A simplified custom object detection model, implemented in a Colab notebook.
An image captioning model, implemented in a Colab notebook.
Full-stack contribution (design, backend, frontend, documentation, testing) to a team-built MVP for a farmer-to-consumer marketplace app - Flutter mobile frontend, Fastify/Node.js backend with Sequelize and JWT auth.
Camera-based perception layer for a mecanum-wheel TurtleBot on ROS2, migrated from a Raspberry Pi 5 to an NVIDIA Jetson Orin for more onboard compute. Course milestone completed; work ongoing.
Multi-label emotion classification (GoEmotions, DistilBERT) tackling class imbalance and rare-emotion detection through systematic experimentation - oversampling, class-weighted loss, per-label threshold calibration, and confusion analysis. Course requirement completed; now continuing at research level.
Independent practice projects from earlier programming years - hobby builds, a thesis, and coursework kept separate from professional and academic AI work.
Bachelor's thesis: a C++ and Python plugin for Plover, an open-source input method editor commonly used with stenotype machines by stenographers.
A 2D space exploration game prototype built in Java to practice the language - a dynamically expanding map, ship control via rotation and thrust, gravity-based orbital mechanics around planets, enemies, and a planet-scanning mechanic for finding fuel.
An early C++ project built while learning the language - a CLI game where an ASCII stick-figure character moves and shoots ASCII "bug" enemies across the terminal.
A Java coursework project implementing a car service / workshop management application.
An early hobby project. A Sokoban-style puzzle game built from scratch with vanilla HTML5 Canvas and JavaScript, no frameworks or build tools.
An early hobby project. A Snake-style puzzle game built for a university web development course with vanilla HTML, CSS and JavaScript.
A command-line Sokoban puzzle implementation in C++.
A tool built for the Analysis II (Calculus II) coursework to generate practice exams.
If the work needs more than a theme tweak, I can help turn the business problem into a clear technical path.
