Internal systems
The unglamorous software a company runs on: presence, scheduling, inventory, approvals. Built around the process you already have, not the one a SaaS assumes.
OFFLINE-FIRST · POSTGRES · ROW-LEVEL ACCESS
Back to portfolioTaking on projects for Q4 2026
I build the internal systems a business actually runs on.
Contract work, remote across AEST or on site in Brisbane. You talk to the person who writes the code — no agency layer, no forty-page proposal.
Internal tools/Operations dashboards/React & Node/Applied AI/Data pipelines/Integrations/Field apps, offline-first/TypeScript end to end/Next.js/PostgreSQL/Supabase/Data modelling/Process automation/Getting off spreadsheets/Reports nobody assembles by hand/Row-level access control/Auth and permissions/Notifications and web push/Installable PWA/End-to-end tests/REST APIs/Python for data/Docker/Deploy and observability/
Where I'm useful
Web products
Customer-facing applications and interfaces in React and Node — fast, accessible, and maintainable by whoever comes after me.
NEXT.JS · REACT · ACCESSIBILITY · PWA
Applied AI
Models, classification and assistants where they pay for themselves — and an honest answer when they don't. Usually the data pipeline is the actual project.
IN-BROWSER CLASSIFICATION · PIPELINES · PYTHON
How I work
- FIRSTUnderstand the process — on site or on call
- BEFORE ANY CODEData model and security boundaries, in writing
- EVERY WEEKSomething working for you to look at
- AT THE ENDDeployed, documented, handed over running
What I hold to
01
I start with the process
Two days on site or on call, watching how the work actually happens. The software comes after.
02
I ship the boring part first
Data capture before dashboards. A clean entry point solves half of what people call an AI problem.
03
I hand it over running
Documented, deployed, and simple enough that the team keeps using it in month six.
Proof
Case 1 of 1
Torre Ativa
Presence intelligence for the field crews building and maintaining electrical networks.
Client: ELECTRICAL GRID CONTRACTOR·ELECTRICAL GRID
The blind spot
Nobody knew where the crews were until the day was over. Supervisors managed by phone call and memory — reacting to yesterday, never to now.
What I built
An offline-first presence app for the field — two taps, no training — feeding a live supervisor view with alerts. Proactive management replacing the daily post-mortem.
What changed
Real-time operational visibility, faster decisions, preventive safety — and a dataset that had never existed before, now the basis for planning.
Built with: NEXT.JS · TYPESCRIPT · SUPABASE · POSTGRESQL · VERCEL
Impact figures for this system exist but haven't been independently verified, so none are published here. Happy to walk through the architecture and the measurement plan.
Is this a fit?
Worth a conversation if
- Your operation runs on spreadsheets and loose messages, and it has already cost you money.
- You want someone who understands the process before proposing screens.
- You need the system documented and handed over running, not a pretty prototype.
- The people who will use it aren't office staff — they're in the field, the shop or the floor.
Probably not me if
- You want a good-looking brochure site and nothing beyond that.
- You need a team of ten starting tomorrow.
- The deadline is non-negotiable and was set before anyone looked at the process.
Start a conversation
Tell me the problem, not the solution
The first conversation is about how the work happens today — what breaks, who works around it, and what it costs. If I'm not the right person for it, I'll say so and point you somewhere better.