Claude Code developer available across New Zealand. I use Anthropic's AI as a build tool, not a substitute for engineering judgement.
Claude Code writes Firebase Cloud Functions, Flutter CRUD scaffolding, real-time sync logic. I design the data model, audit authentication flows, and implement the business constraints that protect your users. Auckland's SaaS builders, Wellington's gov-tech contractors, Christchurch's agritech labs: I've shipped code for all three.
New Zealand's small market punishes waste. I deliver apps that handle NZD payments, comply with Privacy Act 2020, and scale without burning runway on rewrites.
Fixed scope, single invoice. Payment in NZD or EUR, 50% upfront, 50% at delivery.
Claude Code generates Firestore security rules, listener boilerplate, offline-first sync logic. I design the schema, test conflict resolution, and tune indexes for New Zealand latency to australia-southeast1.
Your users see live updates. Your Firestore bill stays predictable. Read how I combine AI with Firebase.
Claude Code writes Stripe webhook handlers, invoice generation, subscription state machines. I validate PCI-DSS flows, handle failed payment retries, and implement the refund logic your support team will actually use.
Auckland SaaS or Queenstown booking platform: the workflow is the same. Payments work, edge cases are handled, your accountant gets clean exports.
Wellington contracts demand audit trails. Claude Code scaffolds immutable logs, role-based access trees, consent management UIs. I review every permission surface, write the Privacy Act compliance checklist, and document data flows for your legal team.
Code that passes a hostile RFP review and survives a ministerial briefing request six months later.
<p>Claude Code is a compiler that speaks English. It generates Firebase Cloud Functions, Flutter state classes, API adapters faster than any human. But it doesn't understand your business model, your compliance surface, or why users churn after onboarding.</p><p>I use Claude Code to eliminate boilerplate. I spend the saved time on architecture that scales, security that holds under attack, and UX that converts.</p>
$ claude-code "Audit the Firestore rules in firestore.rules.
Find any collection readable without auth.
Fix each rule to require authentication.
Add rate limiting on the leads collection.
Run firebase deploy --only firestore:rules when done."
> Reading firestore.rules...
> Found 2 collections with public read: analytics_events, settings
> Patching rules... Done.
> Deploying... ✓ Rules updated.
| Auckland agency | Upwork dev | Hugo + Claude Code | |
|---|---|---|---|
| NZD 20k MVP delivery time | 8-12 weeks | 6-10 weeks | 2 weeks |
| Code you can refactor later | Yes, verbose | Rarely documented | Yes, typed, tested |
| Privacy Act 2020 audit trail | Extra scope | Not included | Built-in from day one |
| Firebase cost optimisation | Rarely prioritised | Ignored | Baked into architecture |
| Response time (NZ hours) | 1-2 business days | Timezone lottery | Same day, NZST-friendly |
| Who owns the roadmap? | Project manager filter | You write every ticket | Direct collaboration, no intermediary |
Claude Code generates syntactically correct TypeScript and Dart. It doesn't generate good architecture.
I design the module boundaries, enforce immutability where it matters, write the integration tests your CTO will actually read. The scaffolding comes from Claude Code. The engineering decisions come from eight years shipping production apps.
If your CTO finds a bug in my code, I fix it for free within the warranty window. That confidence comes from review discipline, not from hoping the AI got it right.
Claude Code doesn't know Privacy Act 2020 exists. I do.
I implement consent logs, data export endpoints, anonymisation pipelines manually. Claude Code writes the CRUD boilerplate around those flows. You get an app that survives an Office of the Privacy Commissioner audit without a single line of retrofitted compliance theatre.
I refactor existing Flutter apps with Claude Code daily. The AI reads your codebase, suggests null-safety migrations, generates missing tests, rewrites God classes into composable services.
Brownfield projects take longer than greenfield because I spend the first week understanding why your previous dev made certain choices. But the workflow stays the same: Claude Code writes repetitive fixes, I validate architectural impact.
I test every Claude Code output before it ships. Firebase emulator for backend logic, Flutter integration tests for UI flows, Firestore rules unit tests for security surfaces.
If a bug reaches production despite that process, I fix it within four hours during NZ business hours, no additional invoice. My reputation depends on reliability, not on blaming the AI when something breaks.
I integrate Stripe for most projects because it handles NZD natively, has webhook reliability, and doesn't require PCI-DSS certification on your side.
Windcave and POLi integration is possible but adds a week to the scope. Claude Code can't generate those SDKs because the APIs aren't in its training data. I write those connectors manually, with test coverage on refund and chargeback flows.
<p>Book a 30-minute call. I'll audit your idea, estimate scope, and send a fixed-price quote within 48 hours. No sales pitch, no upsell, no multi-phase roadmap you didn't ask for.</p><p>You'll know exactly what you're paying for in NZD before we write a single line of code.</p>
Book a call →