I build end-to-end experiences that connect product interfaces, services, AI tools, and real-world devices.
I bring around 9 years of frontend engineering experience to fullstack and AI application development. My background spans complex product interfaces, component systems, Web3 applications, and cross-team delivery. Today I work across Flutter and web clients, Node.js and Python services, model integrations, and edge devices.
I enjoy turning ambiguous requirements into clear service boundaries and usable flows—then reviewing, debugging, and validating the result across the whole stack.
| Focus | Current work |
|---|---|
| Marek Screen | An independently built, Raspberry Pi 5–based AI touchscreen project. It connects a Flutter interface with Node.js/Fastify business services, a Python/FastAPI voice runtime, local tools, and device services. |
| Agent workflows | Voice and text interactions that route to LLMs, authorized tools, local applications, and user feedback. I work on session boundaries, tool permissions, execution receipts, and controlled memory so an AI response can be tied to an actual action. |
| Currently learning | More rigorous LLM evaluation, retrieval-backed workflows, and durable Agent execution patterns. |
A representative interaction path
🎙️ Voice / text → 🧭 Intent routing → 🧠 Model or local tool → 🔐 Authorization → ✅ Execution result → 🖥️ UI / voice feedback
| Project | What it shows |
|---|---|
| halo-theme-joe2.0 | An open-source Halo blog theme and a long-running frontend project. The repository is archived.  |
| sub-proxy | A Koa + React application with deployment and data persistence. |
| mpu6050-sensor-visualization | Sensor data visualization at the intersection of software and hardware. |
Interfaces and fullstack delivery
AI application and Agent workflows
Edge and device integration
Activity cards are generated by third-party services and may be cached or temporarily unavailable.
Open to conversations about AI applications, fullstack engineering, Agent workflows, and open source.