Fractional CTO & Product Engineer
I build AI-native
products end‑to‑end.
Architecture, design, and engineering — for founders and teams who can't afford the wrong hire.
Solo execution
I build the entire product myself — brand through infrastructure. One person holding the whole system in his head means nothing gets lost in translation.
Team leadership
When scale requires it, I architect the solution and lead the engineering team to ship it. Either way: one point of contact, full ownership.
AI-native by default
AI integration architected from day one — not bolted on later. The gap between companies who do this correctly and those who don't is already widening.
Case Study
Kolt's Table
A seasonal cooking platform — designed, built, and launched entirely solo in six months. Brand identity through production infrastructure. Every decision mine, every standard uncompromised.
AI-native from the ground up. Two user-facing features built on Anthropic Claude: natural language recipe search and The Pantry, an ingredient matching system with semantic expansion, ranked results, and generated shopping lists. Ingredient-aware search with synonym expansion finds recipes by what's actually in them, not just titles. A two-stage AI recipe entry system — Sonnet parses, Opus classifies across 100+ filters — eliminates manual data entry. A seasonal intelligence engine drives the entire site automatically — homepage, produce guides, navigation, and editorial content rotate by month with zero manual updates. Behind the scenes, a 20-job Platform Intelligence network operates as a connected system — jobs trigger and inform each other through tracked causal links, not as independent cron tasks. The system manages content lifecycle with predictive EQS decay projection, scans competitors weekly via Claude with web search, drafts newsletters with one-click Mailchimp campaign creation, scores every AI-proposed change with confidence metrics and SERP previews, streams real-time events via SSE, and generates a weekly intelligence report with health trends and EQS movers. Every AI action flows through a centralized approval queue — human review required before any database write.
React 19 · TypeScript · Node.js · PostgreSQL · Docker · Redis · Anthropic Claude · Google Gemini
Case Study
Inkling
A mental-health screening app that helps people decide whether what they're feeling is worth a professional conversation — built native, shipped to TestFlight, and engineered to fail safe.
AI-native in a high-stakes domain. Two stages of Anthropic Claude do the thinking: the first reads a person's own words and routes them to the right validated screener — or straight to crisis support when the language calls for it; the second turns results into plain-language reflection that points toward care, never a diagnosis. The hard part wasn't the model calls — it was making them safe. I built a 46-profile evaluation suite with a dedicated crisis-detection floor that held through every prompt revision, plus 42 out-of-scope guards against misrouting. The app is accountless, collects no personal information, and keeps everything on-device. Crisis support sits one tap away on every screen.
React Native · Expo · TypeScript · Cloudflare Workers · Anthropic Claude (Opus 4.7)
Now taking on new projects
If you're building something
that needs to be done right —
that's the conversation worth having.
[email protected]