Shipping AI products - and the platforms behind them.
Melbourne-based engineer and technical leader. 20 years shipping web platforms across healthcare, fintech and proptech - and lately, AI products I design, build and ship myself, and AI-assisted delivery embedded across the team I lead.
Where I spend my time
Most AI work stalls between a demo that impresses and a system people actually depend on. Closing that gap is architecture, evaluation, cost, latency and a lot of unglamorous production work. It's the part I find interesting.
From idea to production
Taking AI products from idea to live, with the platform underneath them - data model, auth, infrastructure, evaluation and cost control. I make the architecture calls and I write the code.
Making AI stick in a team
Embedding AI-assisted development into how an engineering team already works - shared Claude Code skills and agents, and MCP servers for the tools people already use - so the gains belong to the whole team, not one enthusiast.
Architecture & trade-offs
Build-vs-buy, model and vendor choice, and where the effort is worth spending. Most of it comes down to knowing what breaks at scale and what quietly costs a fortune.
Products I design, build and ship
I own these end to end - architecture, delivery, and the maintenance afterwards. Most are AI-native, with MCP and LLM integration designed in from the start.
aitrips.io
LiveA map-first trip planner built as the companion an AI assistant needs. Plan by talking to Claude or ChatGPT, and the product holds what assistants are bad at - durable state, spatial reasoning, collaborators, offline travel mode. 37 MCP tools in production, and a web UI that's equally capable on its own.
aitrips.io →Jotcal
LiveA writable family calendar for the kitchen wall. Handwrite "soccer Thu 5pm" onto a 13" e-ink panel and it becomes a real, synced event. Native iOS and Android from one design system - the calendar you glance at, not one you scroll through.
TribeRide
In developmentNative iOS group ridesharing for the people you actually know - families, friends, schools, work pods and small communities coordinating rides among themselves instead of hailing strangers. The wedge: flight-aware airport pickup that auto-fills pickup time and terminal from the real flight arrival. CarPlay surface for active rides.
coming soonNext Departure
Open sourceReal-time public transport departure board for e-ink displays. High-contrast UI tuned for Kindle, with geolocation, multi-stop views and a provider pattern for extensible transit support.
GitHub →Pulled live from my repositories every six hours - the actual work, as it lands.
Hands-on leadership that grows with the company
I've progressively taken on more architectural and leadership scope throughout my career - from owning frontend architecture at Oneview Healthcare and Linkly, to leading a technical team at Cohort Go, to now owning full platform architecture and running a cross-functional engineering group at Urban Rest.
Recent work centres on full-stack platform builds on AWS (Lambda, ECS Fargate, RDS/PostgreSQL), React/Next.js frontends, React Native apps, and standing up CI/CD and DevOps practice. I run SCRUM end to end, lead local and offshore engineers, and make the architecture and trade-off calls that come with being the most senior technical person in the room.
I build with AI tools in both production and personal projects - Claude, Claude Code, MCP and direct API integrations - to ship features and automate the development workflow itself. I care about clear code, things that actually work, and not overcomplicating what doesn't need it.
Full-stack platform architecture
AWS CDK v2, ECS Fargate, Lambda, RDS/PostgreSQL, SQS/SNS - end to end.
Cross-functional leadership
Local & offshore engineers, contractors and agencies, with end-to-end SCRUM ownership.
AI-native development
Claude Code, MCP servers, custom agents & skills, LLM integration in production.
Where I've led & shipped
Lead Developer
Aug 2025 - PresentAccommodation platform with furnished apartments across Australia, New Zealand and the UK, serving corporate clients as well as holiday travellers. Architected and grew a Lambda-based Guesty/Salesforce integration into a full platform with its own data store and API layer. I own architecture across three repos - backend monorepo, Next.js web app, and React Native mobile - on AWS CDK v2 (ECS Fargate, RDS/PostgreSQL, SQS/SNS, GitHub Actions CI/CD). Built an LLM triage pipeline for production errors and an admin dashboard for workflows nobody could previously see into. Now building a real-time data platform, Tableau BI, a smart-lock system, and native iOS & Android apps. Rolled out a guest check-in kiosk running in a hotel lobby, and run SCRUM end to end while embedding AI-assisted workflows (Claude Code, custom agents, skills and MCP) across the team.
ReactJS Subject Matter Expert
Jul 2021 - Jun 2025Led React.js frontend modernisation for a global healthcare platform serving hospitals worldwide. Architected microfrontend systems that let teams scale and deploy independently, and drove performance work keeping critical clinical workflows sub-second. Ran the accessibility community of practice and wrote coding standards adopted across teams.
ReactJS Subject Matter Expert
Apr 2020 - Jun 2021Led UX transformation of a mission-critical payment platform serving Commonwealth Bank, Mastercard and JB Hi-Fi. Established React architecture patterns and development standards for the payment platform's frontend.
Technical Lead
Jul 2018 - Mar 2020Led the technical team delivering multi-currency payment services for international education, integrating third-party payment providers. Worked with the CTO on payment-system architecture, ran Agile delivery and mentored junior developers.
A few builds I'm proud of
QLiTE Mobile Platform
Mission-critical iOS app giving officers mobile access to core policing systems in the field.
Console Cloud SaaS
Multi-tenant property management SaaS with a complete trust-accounting system and a GraphQL API.
AI Production Triage
Sentry and CloudWatch issues are deduplicated in n8n, then an LLM works out the likely root cause and severity.
Issue-to-PR Agent Pipeline
A labelled GitHub issue is planned, built and tested by Claude agents in isolated containers, then checked by a review panel before a PR opens. The agents never hold write access.