I designed, built, and maintain this site myself — the same way I run an office: clear systems, small careful changes, and nothing shipped that I haven't reviewed. It's also a working sample of how I use AI: the tools do the labor, and I keep the judgment.
Plain HTML, CSS, and a little vanilla JavaScript, served by GitHub Pages at tanyamlawson.com. No frameworks in production, no cookies, no trackers — pages load fast because there's nothing extra to load.
Libre Baskerville for headlines, IBM Plex Sans for text, JetBrains Mono for labels — on a navy, sage, and gold palette. The 3D glass cards, scroll reveals, and animations are hand-tuned CSS, and every animation respects reduced-motion settings.
The design started in Lovable, was exported to clean static HTML, then refined section by section using Claude Code — every change reviewed in a live preview before shipping. The full history is public in the site's GitHub repository.
AI tools draft, accelerate, and catch mistakes — but nothing goes live without my review and approval. It's the same guardrailed process I bring to operations work, and the same idea behind everything I build: proof before trust.