AI Engineer / Melbourne, Australia
I build production AI products, not demos.
I take products from a hard problem to something real people use — like bydhome.com.au and an RFQ parser running inside a manufacturer.
Selected work
Real products, real users, real consequences
bydhome — a real-estate marketplace, from zero to live
A live, co-founded property marketplace I helped take from a hard problem to something real people use. Equity partner; I own the AI and a good part of the product.
TriCab RFQ parser — unstructured quotes, made structured
A production AI system inside a cable manufacturer that turns unstructured request-for-quote input into structured, quotable data. Client work — shown as workflow and architecture only, no company data.
Writing
Notes from the build
Hybrid routing: cutting LLM cost ~90% without losing accuracy
Sending every query to a model is accurate and expensive. Sending everything to rules is cheap and brittle. Here's the routing pattern I used to get most of both.
Making model output trustworthy enough to show a customer
The gap between a working extraction demo and a system a business will put in front of a customer is almost entirely validation. Here's how I close it.
Evaluating LLM outputs with ROUGE, BLEU and BERTScore
You can't ship AI responsibly if you can't say whether it got better. A practical look at the metrics I used to evaluate model output across 20k+ samples.
Want to talk?