From complex
problems to
working products.

Founder / Product Builder · CTO & Software Engineer · 2025–2026 · Private AI in the browser

A browser-focused local AI product combining inference, retrieval, and interaction with information.

Connects product ownership, browser architecture, applied AI, retrieval, and practical UX under device constraints.

• Product work includes browser-context retrieval, local chat, page summarization, explanation, translation, rewriting, and document questions.
• Developed memory and retrieval experiences around browsing history and connected information.
• Architecture includes a Chrome extension, side panel, background service worker, offscreen execution, content extraction, and IndexedDB.
• Worked with WebLLM/MLC and WebGPU browser inference, local embeddings, and retrieval.
• Worked on document vaults with PDF, Markdown, and DOCX workflows.
• These are areas of product work, not a claim that every historical feature is in the current release. Inference and personal retrieval work run locally; the feed has used server-side collection with local ranking.
• The February 2026 CV records Angular and JavaScript in the Chrome extension, with Python / FastAPI, SQLite, PostgreSQL, and Docker in the wider implementation.
• Built an optimized local retrieval-augmented generation workflow for querying browsing history, generating embeddings, reranking, prompt engineering, and extracting context and data.
• Models used included Qwen 3 4B 2507, Llama 3.2 1B, Qwen3 1.7B, Qwen3 0.6B, and Qwen3 30B, alongside testing other models.

Product Builder · 2026 · Collaborative AI music

A collaborative AI music experience where 2–5 friends shape one original song together, one choice at a time.

Connects a playful multiplayer product with browser-based music generation and a choice of local or cloud execution.

• Built an experience for making AI-generated music together with friends.
• Explored local hardware and cloud execution options.
• Connected model capabilities with a shared creative activity.
• Players join a room, take creative roles, and pass an idea around to shape a shared track without needing musical skills.
• A computer can host generation for other players without taking a player spot; the first person joining becomes the room host.
• The site offers free local AI music generation and an advanced cloud option using prepaid song credits.
• SongCircle links to ai-music-js, the open-source browser music-generation library, so developers can experiment with the technology themselves.

Browser-integrated Local AI Infrastructure · 2026 · Development & product architecture

A browser and engine initiative designed to let ordinary websites request AI models running on users’ hardware.

Explores platform architecture and developer experience at the boundary between web applications and local compute.

• Architecture direction combines a Chromium/Electron browser, a local Python engine, and a developer-facing JavaScript SDK.
• Design work covers permissions, model selection and compatibility, download/cache lifecycle, model sessions, resource visibility, streaming, cancellation, and binary artifacts.
• Intended use cases include transcription, chat, and music generation.
• This is development work and product architecture. The specification describes design intent, not a claim that every capability is shipped or publicly released.

Architect / Developer & Project Manager · 2017–2022 · Conversational-AI product engineering

A drag-and-drop platform for creating complex chatbots without writing code, built with a seven-person team at Chatbotech.

Combined architecture, development, and project management on a flowchart-based chatbot builder.

• A seven-person team created Chatbotech’s drag-and-drop chatbot-development platform based on flowcharts.
• The product was deployed at the National Bank of Slovakia, Erste Group, Tatra banka, multiple shopping centres including Eurovea, e-shops, and insurance companies.
• The platform enabled users to build complex chatbots without needing coding skills.
• My recorded responsibilities spanned architecture, development, and project management. Technologies: Node.js, Express.js, Java, Python, and Angular.

CEO · Past venture · Bioplastic filaments for 3D printing

As CEO, I helped build a company producing 3D-printer filaments from bioplastics, together with a three-person team.

The team sold approximately 1,000 products during the company’s first year.

• A three-person team created the company and brought bioplastic filament products to market for 3D printers.
• As CEO, I handled communication with the technology supplier, hiring, operations, product development, and the supply chain.
• Our team sold approximately 1,000 products in the first year.

Founding-team Work & Product Discovery · Visual collaboration workspace

Worked on Lumisio as part of the founding team, exploring a visual workspace for team collaboration.

The team’s product discovery involved research and interviews with more than 100 people.

• The team explored a visual workspace addressing productivity and collaboration.
• Research and interviews involved more than 100 startup founders, researchers, consultants, and big-tech employees.
• Ongoing conversations informed what the team should build next.

IT Consulting & Software Engineering · IT consulting services

A small IT consulting company through which I primarily provide my own services.

A business base for my consulting and hands-on engineering work.

• Through AITea, I offer my services to clients, with my own consulting and engineering work as the company’s primary focus.

Browser AI & Developer Tools · 2026 · Open-source music-generation library

An open-source JavaScript library for generating instrumental and vocal music locally in a desktop browser.

Brings ACE-Step 1.5 XL Turbo to the browser with ONNX Runtime Web and WebGPU.

• Runs model loading, inference, audio decoding, and WAV creation in a Web Worker, without a generation server or API key.
• Exposes generation progress, cancellation, model caching, and audio output through a developer API.
• Linked from SongCircle for developers who want to run and explore browser music generation themselves.

Experimental; requires a compatible desktop Chrome or Edge setup with WebGPU, sufficient GPU memory, and substantial initial model downloads.

Developer · From 2025 · State synchronization

Developed a TypeScript library for sharing state and messages across multiple devices using operational transformation.

Worked on a reusable synchronization layer for cross-device application state.

• Created a WebSocket library for state and message sharing across devices.
• Used operational transformation to support state synchronization.
• The library was also used in the location-based messaging app. Development was recorded as ongoing in the February 2026 CV.

Independent Engineering · 2026 · Model evaluation & integration

A benchmark experience for evaluating local and cloud AI approaches in practical workflows.

Investigated latency, time to first token, throughput, retrieval quality, interaction costs, and hardware/model trade-offs.

• Built a benchmark experience and compared local inference engines in practical workflows.
• Explored latency, time to first token, throughput, retrieval quality, and interaction costs.
• Self-reported engineering project: ported an ACE music-generation model to Transformers in four days using AI-assisted development. This does not establish an upstream contribution or merged pull request.

Architect / Developer · From 2025 · Mobile messaging

Developed a location-based messaging app for iOS and Android using a state-sharing library I created.

Connected cross-device state sharing with a mobile application across iOS and Android.

• Worked on a location-based messaging mobile app for iOS and Android.
• Built on my WebSocket state library for cross-device state and messaging.
• Technologies: Ionic, Angular, NestJS, and Node.js. Development was recorded as ongoing in the February 2026 CV.

Developer · From 2025 · AI-assisted development & marketing experiment

Built a birth-date-based compatibility website to explore AI-assisted development and audience targeting.

Combined AI-assisted feature and design work with manually implemented payment processing.

• Created a React and Next.js website that presented compatibility based on birth dates as an experiment in an esoteric audience and its marketing.
• Used AI-assisted development for the compatibility feature and part of the design; also implemented design elements manually.
• Implemented payment processing manually. Stack: React, Next.js, and Node.js. Work was recorded as ongoing in the February 2026 CV.

Developer · 2024 · Market-data & machine-learning experiment

Built a bot that collected real-time exchange data, searched for arbitrage, and explored short-horizon price prediction.

Combined streaming market data, time-series storage, monitoring, and machine-learning experimentation.

• Collected real-time data from different exchanges and searched for arbitrage opportunities.
• Experimented with machine-learning predictions over a 15-minute horizon.
• The work exposed data-stream quality and model-tuning constraints when considering scale.
• Technologies: TypeScript, Node.js, Python, Keras, PyTorch, QuestDB, InfluxDB, Prometheus, and Grafana.

Architect / Full-stack Developer · Team Lead · 2020–2021 · Financial application software

Led a three-person development team building an application for automatic invoice processing with NLP data analysis.

Connected team leadership with architecture, full-stack delivery, and language-based data processing.

• Led a development team of three people creating a financial application.
• Built automatic invoice-processing functionality with NLP-based data analysis.
• Worked across architecture and full-stack development using NestJS, Angular, and Python.

Developer · 2016 · University robotics project

Created a four-legged robot with multiple movement patterns and live video streaming to a web control panel.

Connected physical movement and live video with browser-based control.

• Built a quadruped robot as a university project.
• Implemented support for different types of movement and live-streamed video to a web-based control panel.

Developer · 2015 · Mobile game development

Created a game for iOS and Android.

The CV records publication on Google Play and the Apple App Store.

• Created a game for mobile devices on iOS and Android.
• The source CV records app-store availability; no current download destination is asserted.

Independent Development · Web & app project

Worked on a reviewing platform in my spare time, including app development.

Extended independent product development into a reviewing platform and its app experience.

• The original CV describes spare-time development of a reviewing platform, including an app.

Architect / Full-stack Developer · 2018–2019 · Accessibility

Built an accessibility-focused voice translator with an organization working with people with Down syndrome.

Connected machine learning, IoT, and application development to support everyday tasks.

• Worked with an organization focused on Down syndrome to create a voice translator supporting people with disabilities in everyday tasks.
• Contributed architecture and full-stack development using machine learning, Node.js, Python, IoT, and Angular.
• The project also involved hardware development.

Research & Technical Projects · Ontologies & graph systems

Earlier technical work spanning language models, ontologies, and graph-based systems.

A foundation in language processing and graph theory that predates modern generative-AI products.

• Worked with language models and ontologies before the current generation of LLM products.
• Explored graph-based systems and graph-theory work.
• Historical language-processing work is not presented as modern transformer or LLM research.