Publication information

Reuse & credits.

Linked publications

The writing index links to the original published articles. Those editions retain their own terms and credits. The index does not reproduce the article text, illustrations, slides or PDFs.

Website material

© 2026 Vitalii Oborskyi. No general reuse license has been granted for this website's original text or design. Please contact me about reuse. Individual publications and repository components may carry their own licenses.

Author portrait

The author supplied the source photograph for this site. The displayed portrait is an AI-assisted background-removal derivative, optimized as a transparent WebP image. It is not claimed to preserve every foreground pixel. No general reuse permission for the photograph is granted here; contact me before using it elsewhere.

Language labels

The website interface is English. Publication-language labels use EN for English and UA for Ukrainian; machine-readable language codes are en and uk. UA in a project name means Uncertainty Architecture. A platform edition is not automatically a translation; unverified talk languages are omitted.

Display preferences

Auto uses your device's local time: Day from 07:00 until 19:00, Night otherwise. You can choose either theme yourself. Theme and motion preferences are stored locally in your browser; no location request, account or analytics is involved. The decorative spatial background follows native scrolling on each page, and topic choices in Writing. The thematic structures also move gently in a repeating cycle while you read. Motion off freezes both the camera and the structures; your reduced-motion preference overrides a saved On choice. Theme changes only recolor that view. No measurement or scientific simulation is represented. Without JavaScript, your device's light/dark preference and a static geometric background supply the fallback.

Research and acknowledgements

Uncertainty Architecture and The Subprime Code Crisis maintain their own research, evidence and attribution records. The acknowledgements identify UA project advisory roles, specific contributions and public discussions. These relationships do not imply institutional endorsement, certification or adoption by PMI, the University of Waterloo, EPAM or another institution.

Built through AI-assisted delivery

This website is a practical experiment in AI-assisted software delivery. I direct its architecture, requirements, design decisions and acceptance; AI tools support implementation, analysis and verification.

The project explores a question central to my research: how to benefit from faster AI-assisted implementation without losing engineering control, maintainability or accountability.

Architecture and fallback. Content and templates are separate from the rendering engine and visual effects. Generated HTML keeps the site readable without JavaScript; navigation and Canvas are enhancements. Source separation · Fallback checks

Changes and acceptance. Work starts with an issue and explicit acceptance criteria, then proceeds through a linked PR, code review and checks. Test results provide evidence; review, merge and release remain human decisions. Reviewed change · Acceptance boundary

Checks and releases. PR smoke, bounded staging regression and full production regression have distinct profiles. Release checks bind evidence to the exact source and built artifact; staging success does not authorize production. Check profiles · Source and artifact checks

A visual tradeoff. The colored ribbons were removed while retaining the fractal scene, navigation and readable content. This scoped change passed the existing bounded staging checks without relaxing their budgets. Decision and results

This is a personal engineering example of human-directed delivery, with relevance to larger programs. It does not establish enterprise-scale effectiveness, an implementation of Uncertainty Architecture or scientific validation of the research.

Corrections and contact

If you spot an incorrect link, date or attribution, please open an issue or contact me on LinkedIn.