AVAILABLE · Q3 2026
Production generative AI, built end-to-end and shipped.
I’m an independent builder for founders, startups, and companies shipping generative AI products into production. The work I do best is 0 to 1: turning an idea or a vibe-coded prototype into a real, production-quality full-stack app with deep generative AI inside it, in a matter of weeks.
01WHAT I DO
The right generative AI for the job, shipped reliably in production.
The right answer might be a retrieval layer with rigorous evaluations (evals), a multi-step workflow that orchestrates several model calls, or a tool-using agent built on a production-grade harness like the Claude Agent SDK. The harder work is knowing which one actually fits.
01 — GREENFIELD BUILDS
Idea to production app
A discovery sprint to map architecture, evals, and milestones, followed by a fixed-fee build over 6 to 12 weeks. The AI lives inside a real application: frontend, APIs, data, auth, deployment.
02 — PROTOTYPE TO PRODUCTION
Vibe-coded to production
A discovery sprint to map the existing code, decide what to keep, rebuild, or replace, and produce the architecture and eval plan, followed by a fixed-fee build over 6 to 12 weeks.
02HOW I WORK
Scope is named and bounded. Every engagement.
I work AI-assisted throughout, which is how a solo engagement ships at the pace of a small team. Slack within working hours, async-first; on-site visits as the engagement requires.
+ IN SCOPE
− OUT OF SCOPE
→DELIVERABLES AT HANDOFF
03RATE
Both tracks use the same pricing.
Build pricing scales with scope. The lower end is a focused build, typically a single integrated AI feature or a small full-stack app; the upper end involves multiple components, compliance overhead, or scope across several user surfaces.
The discovery sprint produces a fixed-fee quote anchored to your specific scope, plus the architecture and eval plan needed to ship it. Builds are paid per milestone. Net 30 across all engagements.
04FIT
Who I work with — and who I don’t.
+ GOOD FIT IF YOU ARE…
- A founder or startup with a rough or vibe-coded prototype that's interesting but not ready for real users
- A team building a tool-using agent on Claude Agent SDK or similar, who needs a senior engineer to ship it with proper evals and observability
- A non-technical founder who wants the v1 built and delivered
- A funded startup (seed to Series B) adding generative AI to your product for real, not as a demo or a press release
- A team strong on its stack but new to production generative AI
- A venture studio or VC funding the first production AI build for a portfolio company
− NOT A GOOD FIT IF YOU ARE…
- Pre-funding and looking for a co-founder for equity
- Looking for someone to write strategy decks
- Convinced you need an autonomous agent before you've defined the problem
- Looking for an embedded engineer who works inside your team's process day to day
05BACKGROUND
Two decades of engineering and product leadership.
Inventor on US12380340B1, Systems and Methods for AI Agent Control Layer Through Performance Evaluation of Artificial Intelligence Agents.
Wayfound
Pendo.io
Box
Zenph · EvoApp · Siemens Healthcare
06GET IN TOUCH
Tell me what you’re building.
The first reply is from me, usually within a working day. Or skip ahead and book a 30-minute call directly.