Agentic MCP Teaming
Multi-agent coordination framework where named specialists collaborate via a shared MCP bus: parallel review, unanimous consensus on design, and isolated Git worktree implementation.
View project →Solutions engineering leader with 15+ years helping enterprises and governments across ANZ, ASEAN and Greater China adopt emerging technology with confidence. I have built and scaled technical pre-sales practices from the ground up at every major infrastructure shift (security, data protection, cloud and cloud-native), translating complex platforms into evaluations that risk-averse buyers will actually deploy. Today I am focused on GenAI: embedding it into how my global team solves customer problems, and on the trust, security and identity questions that decide whether enterprises move AI into production.
Projects spanning infrastructure automation, security platforms, and agentic tooling, built to solve real operational problems.
Multi-agent coordination framework where named specialists collaborate via a shared MCP bus: parallel review, unanimous consensus on design, and isolated Git worktree implementation.
View project →A harness-agnostic control platform for agent coding sessions: compounding Markdown memory with local hybrid retrieval, a curation loop that keeps it true, and an always-on guidance and execution boundary every session inherits.
View project →Scheduled LLM risk verdicts for Renovate dependency PRs: a fail-closed automerge gate where the model can only withhold approval, never grant it without deterministic CI passing first.
View project →An autonomous agent that operates the homelab end to end, proposing changes, running deployments, and keeping a durable memory of past decisions so the why survives between sessions, under its own least-privilege identity.
View project →Zone-based firewall policy framework with a Go/React editor, deterministic OpenTofu compiler, compile-time break-glass invariants, and advisory LLM risk evaluation before apply.
View project →Production Go service that correlates infrastructure from Proxmox, DNS, and UniFi into unified NetBox entries: multi-source deduplication, continuous sync, and Prometheus metrics for full visibility.
View project →Sysdig
Lead technical pre-sales across ANZ, ASEAN and Greater China for cloud-native security, focused on large enterprise, high-security and financial-services accounts.
Tenable
First dedicated Cloud Security SE for Tenable in APJ; built the cloud security go-to-market through acquisition integration (Ermetic), with a heavy focus on identity.
Tenable
Developed the Northern ANZ market (NSW, ACT, NT, QLD), focused on financial services and government.
Redlands
Interim ICT leadership with overall responsibility for the technology platform, architecture and infrastructure projects.
Apple Inc.
Pre-sales systems engineering for the NSW education market, spanning K–12 through higher education.
RSA, Security Division of EMC
Broadened remit across DLP, network security analytics and governance tooling; helped rebuild the RSA brand following the RSA breach.
Symantec
Security pre-sales for NSW, growing from mid-market into large enterprise.
Alphawest Services (Optus Business) · KineticIT
Blending executive stakeholder management with deep technical credibility across security, cloud, and emerging AI platforms.
Bachelor of Computer and Mathematical Science
University of Western Australia · 2003
Majors: Information Technology Systems, Information Technology Applications, Electronic Commerce.
Notes on solutions engineering, platform security, and practical GenAI.
My agent's memory got very good at knowing what was true and had no idea what any of it was in aid of. So I gave it a statement of ends, a telos, and a sweep that asks which of my goals currently have nothing moving. The interesting part wasn't the machinery. It was having to write the goals down.
Read postA while ago I explained that my AI agent's memory is just a folder of Markdown notes, the same note-taking stack I use for myself. Then the folder got big. This is about the three things a notebook needs once it stops being small, and how each one turned out to be something a human brain already does.
Read postMy AI agent keeps a memory between sessions. Everyone assumes that means a vector database and an API bill. It's actually Obsidian, Markdown, and a local search engine, the same note-taking stack I use for myself. Here's why human and machine memory quietly converged on the same design.
Read post