Selected work

Systems built,
results shipped.

Every project here started as a manual process. Each one now runs itself — in production at Wego, daily. Built with n8n, Claude, and a bias for what’s measurable.

AI Operations★ Flagship

Live-Blog WordPress Plugin

Built in a weekend during the US–Iran travel crisis — a custom GPT wrote the PRD, Claude wrote the plugin in one session. Three production live blogs run on it today; the MENA travel-disruption page held #1 on Wego EN and AR for 3+ months.

WordPressClaude AICustom GPTn8nEN + AR
Read the full story
EN sessions846,000
combined EN + AR~1,000,000
rank held#1 · 3+ months
build timeone weekend
SEO & Growth

Organic Growth Sprint

Flat traffic, no signal on what to prioritise. Connected GSC + GA4 into a daily n8n pipeline surfacing high-impression / low-CTR pages, briefed by Claude — one cluster, one owner, no paid media.

+230%sessions21,100+weekly peak10 wkssprint
GSC APIGA4 APIClaude AIn8n
Read case study
Content Automation

Bilingual CMS Pipeline

A Claude-driven publishing skill that drafts EN + AR variants from one brief and writes straight into Prismic’s slice schema — the real structured model, not a flat blob.

30+pages / runEN+ARone trigger0manual steps
n8nClaude AIPrismic CMS
Read case study
AI Operations

Airline Bulk Update System

Updating schedules, routes and copy for 34 airlines used to be 1–2 hours of manual WordPress editing per cycle. Now: structured sheet → Python transform → parallel REST pushes, with error logging and a Slack digest.

34airlines / run~10 minwas 1–2 hrs0data-entry errors
Pythonn8nWordPress REST API
Localisation

App Store Screenshot Automation

One master Figma file as the single source of truth: Claude adapts tone and phrasing per market, Figma MCP writes localised copy back into each language frame, and the export pipeline delivers upload-ready assets for every store locale in one run.

1master fileeverystore locale1trigger
Figma MCPClaude AIElevenLabsn8n

Methodology

How I work.

Every automation follows the same four-step discipline — from finding the bottleneck to proving the outcome with a number.

01
Audit the bottleneck
Tasks taking 30+ minutes on a repeatable pattern. If a person does the same thing more than twice, there’s a system waiting to be built.
02
Architect the flow
Every step, API call and decision point mapped before a line is written. The n8n diagram comes first; implementation second.
03
Build and deploy
Everything runs in production. No demo environments. If it doesn’t run daily without supervision, it isn’t done.
04
Measure the delta
Before and after — sessions, publish times, error rates. The number is the proof, not the workflow screenshot.

Currently available

Open to AI automation lead, AI operations & content automation roles.

AustraliaNew ZealandEUCanadaRemote