AI Automation Developer · Adelaide
n8n workflows, RAG chatbots, and lead systems — the boring machinery that actually ships to production. Currently building the AI systems for a seven-brand travel group.
Selected work
Every project starts with the problem, ends with a measured result. Client system names kept private.
Lead automation
What I built: a daily automation that reconciles every website enquiry against the CRM and flags the ones that never arrived — the gap between the form and the system where leads quietly vanish.
Why it works: integrations fail silently — no error, no alert. The job of the automation is to watch the gap everyone assumes is fine.
Recovered lost leads across 7 brand sites
RAG chatbot
What I built: assistants that answer only from the business's own documents — and say "I don't know, let me get a human" instead of inventing an answer. Backed by a 90+ case test suite before anything reaches a customer.
Why it works: a confident wrong answer isn't a bug, it's a liability. Boring and correct beats clever and wrong.
−20% hallucinations, measured 50+ client knowledge bases
AI agents
What I built: an AI agent on the site that answers FAQs instantly and decides the next action — save a lead, open a ticket, or book an appointment automatically.
Why it works: the win isn't a smarter chatbot, it's fewer humans doing manual triage. The site starts earning its keep.
+25% client engagement −40% manual booking time
What I believe
Hype is the tell of someone who's never shipped. I'd rather show you the scar than the highlight reel — because the scar is where the real engineering is.
Real numbers beat vibes. No metric means it didn't really happen.
Showing what broke beats showing the win. The failure is where you learn whether it's real.
You don't have to be finished to do real work. I shipped production AI before I graduated.
How I ship
Over half of GenAI projects get abandoned after the pilot (Gartner). The difference isn't a better model — it's the boring machinery around it. Here's mine.
The bot only knows what the business knows. Scheduled scrapes keep it current — stale knowledge is how bots start improvising.
No source, no answer. If retrieval comes back empty it hands off to a human and saves the lead — never guesses.
90+ test cases across 16 categories — prompt injection, strict date rules, honest "I don't know"s — re-run after every change. AI interrogates the bot before any human sees it.
The team tests it first — every bug logged with a launch gate — then a small customer beta. I read the real conversations daily and tune to the brand.
Work with me
Freelance builds, consulting, or a full-time role. Usually reply within a day.
I run a cross-check on your forms for a week and hand you a report of the actual leads slipping between your website and your CRM. The recovered leads pay for it.
fixed scope · fastA RAG assistant that answers only from your real docs, hands off when it doesn't know, and ships with a full test suite. The version you're proud of, not liable for.
build + handoverOngoing n8n workflows that watch the gaps everyone assumes are fine — reconciliations, alerts, the quiet failures. I keep the machinery honest.
monthlySignal
Placeholder quotes — swap for real client/colleague words before launch.
“She shipped the thing that actually worked in production — not a demo. That's rarer than it should be.”
— placeholder, replace with a real quote
“Caught leads we didn't even know we were losing. Paid for itself the first week.”
— placeholder, replace with a real quote
About
i'm Madelynn — an AI software developer in Adelaide. i shipped production AI for real businesses before i'd even graduated, and i've been doing it for three years across three companies and 60+ clients.
my thing is the guardrails, the test suites, the watching-the-gaps work that keeps an AI system honest once it's live. most AI projects die between the demo and production. i build the ones that don't.
Bachelor of Computer Science (AI major), University of Adelaide — GPA 6.8/7.0.
Let's talk
Freelance automation builds, AI consulting, or a full-time role.