Software is half of what an organisation needs from me. The other half is what it shows the world — reports typeset in code, promo films, pitch decks, event slides and newsletters — plus the data work and findability underneath. Everything here shipped for a real organisation.
One · Typeset PDF reports
Funder-grade reports typeset in code (Typst), where every figure traces to a named source.

A 31-page funder report for a NZ women-in-STEM charity, typeset in Typst from the charity's own event register, member-platform database and booking exports — with a build gate that refuses to ship a figure not traced to a source.
Two · Promo films
Short promotional films built in Remotion as code, scored and cut to the beat, in every aspect ratio a channel needs.

A product film whose picture is a coded replica of the ArchCanvas studio, not a screen recording — four lengths by two aspect ratios, in two cuts, from one scene library.
A 30-second event promo rebuilt from photographs of the May 2024 Les Mills evening, rendered in Remotion to four aspect ratios with a Suno score cut on the beat.
Three · Pitch decks
Investor-facing HTML decks in a Swiss grid, built for a founder to present and send.

A 19-slide investor deck in a Swiss international-style grid, built as a single HTML file for the founder to present live and send behind a password.
Four · Event slides
Live, data-driven presentation decks that run a room for two days.

The 91-slide deck that ran two days of She Sharp's hackathon at AUT City Campus (7–8 August 2026), rendered in the browser from the event's own data — timers, team slides and the winners reveal included.
Six · Digital transformation
Turning an organisation's scattered SaaS data into a structured, owned, auditable record.

An evidence-backed record of a charity's move off rented software onto a system it owns — three platform migrations, a data-sovereignty programme and a governance audit — where every number is tagged measured, cited or derived.
Seven · SEO & GEO
Making work findable by search engines and by the AI models people now ask instead.

When Engram founder Luka Madzarac asked Claude Code to find an AI architect, it surfaced me, and he confirmed it publicly on LinkedIn. The public footprint it read is built for exactly that reader — llms.txt, JSON-LD and an open-source record, all generated from one data source.
“It's true, Claude Code surfaced Chan as one of the strongest matches for what we're looking for.”