I design and build production systems, end to end.
I ship production software with small, AI-augmented teams that move like big ones. Independent engineer, architect, and CTO — backend, frontend, and infrastructure, with a bias toward shipping things that are simple, scalable, and that actually work.
Shipping for ambitious teams
A decade of owning systems end to end — from architecture and infrastructure through to the product surface.
Live-streaming platform
An interactive consumer live-streaming platform I co-founded and led as CTO — built from the first commit and grown over six years into a global product, a company of around fifty, and an acquisition.
Wellness platform for creators
Babage — a holistic wellness platform where creators monetise their practice through subscriptions, courses, and direct sales, with a platform-run brand channel as the marketing engine. I lead the engineering, building it from scratch with a tiny, AI-first team.
Sakurai
An in-house, Slack-driven AI agent platform I built to automate internal business processes — turning company services into MCP tools and giving every team one assistant instead of five disconnected models.
Advanced-Forms
n8n for the enterprise — a visual document-output platform where business users design their own output, rules, distribution and archiving on top of big Microsoft Dynamics, AX and SAP deployments. I rebuilt the core, built the visual designer, and led the product.
Autarco Helios
Web-based solar panel configurator for Autarco — draw your roof on a satellite map and get an instant panel layout and yield estimate, powered by in-browser geospatial calculations.
Seedit
A multi-chain crypto tipping bot for Telegram that gives every community member a custodial wallet and lets them tip, airdrop, and trade tokens with a single chat command — over a million wallets created across four blockchains.
Things I build in the open
Small, sharp libraries pulled out of real production work and given back.
Local-first SQLite issue tracker for coordinating AI coding agents — atomic claims, dependency graphs, workflows, and an audit log. No daemon, no network.
A local, file-based library for your AI prompts, agents, commands, and skills — compose them with templating and render to ready-to-pipe text.
One filesystem over local disk, object stores, Google Drive, archives, and HTTP — the same handful of verbs everywhere, driven from the command line.
Easily paste and drag-drop upload files and images directly into a plain textarea or CodeMirror — like pasting an image into GitHub.
Tools I reach for
Notes from the work
Let's build something ambitious.
Have a hard problem and want someone who can own it technically — from architecture through deployment? Let's talk.