Software Architect& Engineer
Designing and building software is what I enjoy most. I've done it on embedded systems, on large web platforms and lately with AI on the device — as architect, as team lead, and always hands-on.
A few numbers along the way
2M
pages on a job search platform, server-rendered with Next.js and all-green Web Vitals
40k
seats at 60 fps on a phone, drawn with WebGL for a ticketing platform
128 KB
of RAM — enough for a browser UI and data streaming on a measurement device
2.7k
GitHub stars on two small open-source tools, lua-websockets and login-with
Domains & industries (06)
2026 –
On-device AI
Local LLM inference on unified-memory hardware, agents that work without a network, in-browser inference with WebGPU and WebNN
2023 – 26
Ticketing
A WebGL seat map for 40,000 seats, a Kafka work queue for sales peaks, sales dashboards on ClickHouse
2019 – 23
Job search
Two million server-rendered pages, layered caching, all-green Web Vitals, a frontend team of three to five
2008 – 18
Test and measurement
Kernel drivers, protocols and data streaming, web tech on devices with 128 KB of RAM, the browser as instrument panel
2005 – 08
Industrial vision
Realtime C++ for multi-camera inspection on pharma lines, up to 100 inspections per second
2012 –
Open source
Small libraries and services like lua-websockets and login-with, used more than expected