// about
I'm a senior software engineer from the Czech Republic with a master's in computer science and 8+ years of experience. My neural networks scanned 6+ million images in the fight against child abuse material, and my security research made national news.
At Boost.space I build on vector search and LLMs, develop the MCP server and work on the PHP/Nette API.
After hours: model-level computer vision, training and inference, LLM and cybersecurity research.
// stack
// Highlights
AI forensics against child abuse material
Deep-learning research with real-world impact
Built a CNN-based forensic tool (COPINE scale) that analyzed 6+ million photos on Rajče.net. Covered by tech media, awarded by the Governor of the Zlín Region.
Flaw in the national vaccination registry
Security research · CNN Prima News
Demonstrated how the Czech COVID-19 vaccination booking system could be bypassed.
Open source
Contributte · php-eet · CVE-2020-15227 checker
1,500+ GitHub contributions in the last year.
// Experience
Backend Engineer
Boost.space
Building the platform's MCP server end-to-end, plus vector search and retrieval algorithms — working on the edge between LLMs and backend engineering (Python, PHP/Nette).
Tutor & Mentor
Tomas Bata University in Zlín
Tutored programming and neural networks at the Programming Support Center.
Senior PHP Developer
NETVOR · Fondee.cz
Helped design the API architecture for an online investment platform.
Network & Security Specialist
Petra Clinic
Designed and ran a GDPR-compliant network protecting sensitive medical data.
// Education
MSc & BSc in Computer Science
Tomas Bata University in Zlín
Both theses on deep-learning forensics; bachelor thesis awarded by the Governor of the Zlín Region.
// writing
Claude Code: stop burning tokens on the wrong effort level
On FrontierCode, Sonnet 5.5 scores 52.1% at xhigh and 46.2% at max, and the max run costs 13 times more. What effort changes in Claude Code and how to set it.
Classify images with TensorFlow and Deep Java Library
You will rarely find a tutorial that shows how to properly load a TensorFlow Protocol Buffers model into the Deep Java Library.