I architect AI agents, ship full-stack platforms end-to-end, and teach others to do the same.
Open to founding-team / staff-IC / AI architecture engagements starting Q3 2026. Book a call →
What I do
Build AI agents that don't fall over.
Software that decides and acts on its own — with safety rules enforced in code, not just hopeful prompts — so it's safe for money, health data, and compliance steps.
Ship whole products end-to-end.
From database to interface to billing, often solo or as the first engineer — platforms with paying customers and real clients that survive big rewrites.
Keep AI fast and affordable.
Predictable speed and cost through caching, model routing, and tested failure modes — in production, apps that open in under a millisecond.
Teach it.
I designed and taught TechNest's 12-week AI course; 30 students graduated having shipped real AI apps. My rule of thumb: natural language is the source code.
Selected work
Founder & Sole Engineer
Apr 2026 — PresentArchCanvas
Type a brief in plain English and get back a dimensioned floor plan you can actually build from, then refine it by talking. Underneath is ArchLang, the open-source floor-plan language I also wrote, so every edit stays exact instead of being regenerated.
Chief Technology Officer
Mar 2025 — PresentFemTech Weekend
Sole engineer across two platform generations for an organisation self-described as China's first in women's health technology. The current platform carried the 2026 Shanghai Summit: four days, 20 speakers, headlined by Ida Tin, who coined the term FemTech.
Senior Full Stack Engineer & Website Team Lead
Jul 2025 — Sep 2026She Sharp
Recruited by founder Dr Mahsa Mohaghegh, I moved a volunteer-run women-in-STEM charity off rented software onto infrastructure it owns — a new website and a 1,549-person mailing list cut over with 0 failures — and wrote its recurring work down as 11 agent skills a non-engineer can follow.
Founding Principal Engineer, Activation, Execution & AI Systems
Oct 2025 — Sep 2026GAVIGO
Promoted from Core Engineer to Founding Principal Engineer at a Delaware startup. I owned the intelligence layer that let a tapped game open in under a millisecond, and moved the platform from DigitalOcean to Google Cloud in a ~30-minute cutover.
AI Instructor & Mentor
Apr 2026 — Aug 2026TechNest Community
Sole instructor of TechNest's first AI-specialised track — a 12-week, prompt-first course with live classes and an in-course AI assistant. 30 students graduated, and six capstone teams shipped working AI apps.
AI Agent Architect
May 2026 — Jul 2026Engram
Recruited onto Anthropic's partner-architect track after an AI surfaced my open-source work to Engram's founder, who confirmed it publicly. I completed the Partner Network Learning Path and passed the Claude Certified Architect practice exam — credentials behind Engram's approval into the Partner Network.
Earlier — CTO at women's-health startup Sanicle (led a team of five and built the IBM watsonx integration behind its IBM Silver Partner certification) and at period-equity startup FreePeriod, mentor with Forward With Her, backend engineer on ByteDance's Youth Training Camp, and a Lincoln University industry placement building a field-operations app for NZ civil-construction firm CORDE.
Things I've built
Teaching & giving back
Everything I learn how to build, I publish for free.
Five cohorts of curriculum since 2024 — all of it still online, in English and Chinese, no paywall and no signup. Students arrive knowing only how to chat with an AI and leave having shipped a real app that other people use.
Recognition
UN CSW 69 Speaker
United Nations HQ, New York — presented Sanicle AI at the Beyond Beijing 30 conference; drew IBM pilot interest
Outstanding Mentor Award, two years running
Aotearoa AI Hackathon Festival 2026 (125 registered participants, twelve teams; also MC'd the opening) and AI Hackathon Festival 2025 (1 of 14 expert mentors, 11 teams, 80+ people)
a11y-loop — being adopted by My Life My Voice
The disability-led NZ organisation that set the accessible-UI challenge at the 2026 festival is taking up the open-source tool I built for it, for its own development team
UN Women FemTech Hackathon — Outstanding Performer
2025 UN Women FemTech Hackathon, Beijing — recognised as an independent developer
FemTech China Excellence Award
FemTech China Women's Health Technology Challenge — core team member
Master's with Distinction
Master of Applied Computing, Lincoln University NZ — Dean's List, top 5%
What others say
Frequently asked
Who is Chan Meng?
Chan Meng is an AI Agent Architect & Full-stack Engineer based in Auckland, New Zealand. She architects production AI agents, ships full-stack platforms end-to-end — for paying customers, sensitive health data, and regulated work, not demos — and teaches others to do the same. She holds a Master of Applied Computing with Distinction from Lincoln University.
What does Chan Meng build?
She builds AI agents that act safely on their own, with guardrails enforced in code rather than hopeful prompts, and ships whole products from database to interface to billing — often solo or as the first engineer. Recent work includes a Māori-culture AI wellbeing coach, ArchCanvas (an AI design agent for architects, powered by her own ArchLang floor-plan language), and an AI career agent.
What is Chan Meng's tech stack?
Her core stack is AI agents — Claude Code, Model Context Protocol (MCP), LangGraph, CopilotKit, and RAG — with TypeScript, Python, and Go. On the full-stack side she uses React, Next.js, and React Native on the front end, and Node.js, PostgreSQL, Drizzle, Redis, Cloudflare, Docker, and Stripe on the back end.
What are Chan Meng's most notable projects?
Her flagship builds include GAVIGO IRE, a Delaware startup's app-launch platform where tapped games and videos open in under a millisecond; ArchCanvas, an AI design agent that turns a plain-language brief into a buildable floor plan, powered by her own ArchLang language; Tam-AI-Ti, a te reo Māori AI wellbeing coach; and She Sharp, the member platform she led the build of for a volunteer-run New Zealand women-in-STEM charity. She also shipped one of the first Google News MCP servers, 35 days after the standard launched.
Does Chan Meng teach, and is her course material free?
Yes. She designs and solo-teaches AI-programming cohorts — five since 2024 — and publishes the entire curriculum free at programming.chanmeng.org, with no paywall and no signup. All five cohorts stay online permanently, and every one of the 153 lesson pages exists in both English and Chinese. The 2026 TECHNEST track graduated 30 students who shipped six deployed multi-user AI products to a public Demo Day. The platform itself is Apache-2.0 open source and includes an AI tutor that answers from each class's own lessons. She also won the Outstanding Mentor Award at AI Hackathon Festival 2025.
Is Chan Meng available for work or collaboration?
Yes — she's open to founding-team, staff-IC, and AI architecture engagements starting Q3 2026, alongside technical consulting, speaking, and mentorship, with a focus on AI agents, cultural technology, and women's health. Reach her at hello@chanmeng.org or book a 30-minute call at cal.com/chan-meng/30min; she typically responds within 24–48 hours.
Where is Chan Meng based?
Chan Meng is based in Auckland, New Zealand, and works with teams across New Zealand, China, and the remote US. She is bilingual in English and Chinese, and spoke at UN CSW 69 at the United Nations headquarters in New York in March 2025.
The Week in AI, Explained in Five Minutes
Monday mornings: one big story with why it matters, the rest of the week's news grouped by theme, and the new tools worth a look. I write it with Claude Code and read every issue before it sends — every link goes to the original.
Read past issues in the archive, or see how it's made.
Let's build
something
Hiring, collaboration, or AI-tooling advice — email hello@chanmeng.org or grab a time below.