I’ve spent 25 years designing, selling, and operating network and security infrastructure. Now I’m building AI tools for legal practice and studying law at Swinburne. The common thread is solving complex problems for people who need reliable answers.
#The short version
I use Claude Code to build and operate production systems that would traditionally need a dedicated engineering team. The code is AI-assisted. The architecture, operations, and judgment are mine.
That’s what one infrastructure person can do when the bottleneck shifts from typing to thinking.
#Career arc
My career started in Melbourne in 2000 at a small ISP, building foundational skills in Linux systems, routing protocols, and network operations. From there, iiNet (Greater Melbourne Area) — deeper network engineering, larger scale, more complex problems.
Uecomm (later acquired by PIPE Networks) was carrier ethernet: metro fibre deployments where a misconfiguration didn’t just break someone’s internet, it broke a hospital’s connectivity or a bank’s trading floor uplink. The operational discipline from carrier-grade work never left.
O2 Networks (now Telstra-owned) was network and security professional services consulting — a lean team doing high-complexity work for enterprise and carrier clients. Every engagement was a different set of constraints, a different failure mode, a different answer to “what does good look like here.”
Then pre-sales and solutions architecture: NTT (through the Dimension Data acquisition) for enterprise networking at large accounts; Brocade at the front of the SDN and NFV wave; Check Point for security engineering across network, cloud, and endpoint.
Fastly was the last stop before the current work — six months as Senior Security Technical Account Manager for ANZ, at the intersection of edge computing, CDN performance, and application security.
#Current focus
Three legal AI projects, each at a different stage.
auslaw-mcp is live and open. A semantic search tool for Australian case law: plain-language questions, no keyword guessing, returns the actual judgments it retrieved with citations. Built because existing tools were poor. Try it on GitHub.
Silk is in development. An AI workflow platform for barristers, starting with the Victorian Bar. Brief triage, document search within large bundles, citation synthesis, standard form correspondence. Designed around the barrister’s actual working reality: instructed by solicitors, high document volume, deadline pressure, no support staff.
A family law litigation support platform (brand name TBD) grew out of my own experience navigating the family law system. Tooling proven under real filing pressure. Now productising for solicitors, paralegals, and self-represented litigants.
Beyond the legal work: a multi-agent orchestration platform coordinating Claude Code sessions across distributed workers, and the infrastructure that runs everything — 90+ services on a two-node k3s cluster operated by one person.
#Doing an LLB
Studying at Swinburne, targeting the Victorian Bar. The legal AI tools aren’t academic exercises; I built them because I needed them. The intersection of deep technical background and a legal qualification is genuinely rare in Australia, and the cases where it matters — privacy litigation, cybersecurity breaches, disputes involving AI systems, digital forensic evidence — are growing, not shrinking.
#How I work
Claude Code writes the majority of the code. I’m transparent about that because it matters. The architecture, operations, and judgment are mine. Claiming otherwise would be dishonest, and claiming the code is the hard part would be wrong.