Work

Context

Co-founded a decentralised sovereign identity platform on Ethereum: self-custodial wallets, biometric KYC, AML / sanctions / PEP screening, verifiable credentials. Built for the Australian regulatory environment (AUSTRAC, ASIC, Privacy Act) with international rollout in view. Post-quantum-resilient cryptographic stack underneath.

What I did

Architecting and building the production stack end-to-end as lead engineer. Cloudflare Workers backend (Hono, Drizzle ORM on D1, better-auth, SIWE for Ethereum sign-in, Queues for async email delivery), React 19 PWA frontend (wagmi / viem for web3 wallet integration, shadcn / Radix component system), AWS SES + react-email for transactional mail, Sentry observability with kill-switches and structured audit logging for regulatory defensibility. Wrote the launch-readiness reports, compliance documentation and incident runbook. Ran k6 load tests against the authenticated hot paths.

Outcome

Pre-launch, deployed to production and dev environments. Resilience kit, audit foundation and email infrastructure shipped. Closing out the final launch-blocking work.

Context

Ran a six-figure Sydney health coaching business with 40-plus active clients across three years. The bottleneck was always between sessions. Off-the-shelf coaching apps were generic and did not carry my voice, my programming philosophy or the specifics of how my clients actually train.

What I did

Designed and shipped a cross-platform mobile coaching app (iOS and Android) with a Claude-powered conversational coach trained on my own coaching style and content library. Owned the full stack end-to-end: mobile clients, backend API, prompt engineering, Claude integration, deployment, customer onboarding. Iterated weekly on real client feedback.

Outcome

A coach in every client's pocket that sounded like me. Engagement between sessions went up, retention improved, my evenings got quieter.

Context

A growing creative agency where account managers were ferrying data by hand between Monday, email, brief documents and reporting. Monday was the system of record but the humans were doing the system's job. Hours a week, every week, across every account.

What I did

Embedded into the team for a quarter. Mapped the actual end-to-end workflows (not the documented ones), identified the highest-leverage automation targets, then built APIs into the Monday CRM and a set of scripted integrations that handled the repetitive work. Trained the team on the new flow so it kept running after I left.

Outcome

A long list of manual workflows now run themselves. Account managers got their afternoons back for the work they were actually hired to do.

Context

Nuclear bids for major Australian infrastructure programmes needed sharper risk-weighted revenue modelling than the existing tooling supported. Bid risk shapes the entire commercial story, so the precision flows straight through to whether you win and on what terms.

What I did

Built a Python Monte Carlo simulator from scratch to model risk-weighted revenue distributions across live bids. Wired it into existing project controls workflows. Trained the bid leads on reading and challenging the outputs rather than accepting them as gospel.

Outcome

Tighter, defensible risk numbers feeding into competitive bids on multi-million-dollar nuclear infrastructure work. The kind of input a bid director can actually take to a steering committee.

Context

Established performer with a full booking calendar, but a website that wasn't doing the work. Visitors weren't converting into enquiries because the site didn't carry the polish or social proof of the live performance.

What I did

Designed and shipped a polished single-page site with a testimonials carousel, an 8-image gallery, embedded performance video, and a contact form wired to email. End-to-end in a week.

Outcome

Site reflects the actual quality of the work. Enquiries up, friction down.