Young Kiwis in AI: June Meetup

The Young Kiwis in AI meetups give young people working with AI a place to share what they are building and what they have learned along the way. The June 2026 session featured two member spotlights with very different starting points: Lachie Christie, co-founder of GrowLab Technologies, on bringing AI to the infrastructure sector, and Caleb Wharton, a graduate AI engineer at RUSH, on building AI products and helping teams adopt them.

Lachie Christie: AI for infrastructure at GrowLab Technologies

Lachie Christie describes his path as a little different to most in the room. He is not originally from New Zealand and only arrived in Auckland in late 2023, having started out in Sydney. There he studied a combination of mechanical engineering and a Bachelor of Creative Intelligence and Innovation, a degree built around applying design thinking and human-centred design to business problems. It paired a core degree for depth with a transdisciplinary breadth, and was aimed at turning out consultants or entrepreneurs.

He knew he was interested in entrepreneurship but not yet what that looked like, so he fell into consulting at Aurecon, a large engineering, design and advisory firm. Working in asset management advisory, he supported mostly government agencies and public asset owners on how to better manage infrastructure across road, rail, health, water, energy and local government. The work ranged from overseeing divers on Navy bases to writing digital twin strategies from the office.

When the New Zealand team needed to build up its asset management capability, he did fly-in fly-out work for much of 2023 before scoping a role to move over and lead the practice here. That is what brought him to New Zealand, going, as he puts it, in the opposite direction to most. He then convinced one of the great people he met locally to leave and start a business with him.

The business began as something completely different, but they pivoted at the last minute to an AI consultancy and started around March 2025. Lachie splits the journey into exploration, for most of the first year, and traction from this year on, though he keeps an emphasis on exploration because things are changing so quickly and he is still working out where he wants to play.

GrowLab's initial thesis was that while large firms like Aurecon had dedicated generative AI teams, most small to midsize players lacked the resources or appetite to invest, so GrowLab would partner with small to midsize engineering consultancies as an external strategy and development function. That proved harder than expected: a tough economy, general risk aversion in the sector, and the market simply being early on AI maturity all made traction difficult through last year.

His key lesson from that period is to be deliberate about who you work with, both partners and customers, and how much meaning you draw from it. Another is to always push for a seat at the table. For much of his time at Aurecon he was the most junior person in the room, which he sees as a great place to be early in your career: you get the chance to prove yourself and to learn from people with more experience.

GrowLab is now getting traction in infrastructure. Projects range from custom websites with an embedded retrieval-augmented generation agent that qualifies leads through the sales funnel, to a proof of concept for a former government client that uses large language models to take manual work out of a document-heavy process. The AI summarises property documents, pre-populates the forms staff take to site visits, and drafts the final reports for them to review, edit and approve. That is now leading to a larger piece of work, alongside smaller business automation around things like social media and expenses.

Asked what he looks for when hiring, having taken on someone from the Young Kiwis in AI community, Lachie says he has intentionally targeted people with no experience, to get a fresh, AI-first perspective without the assumption that what worked before applies now. He wants people who lean on AI to generate code efficiently but still make the architecture decisions and understand the trade-offs themselves. And with almost anyone now able to build something, he rates the softer skills highly: critical thinking, planning, and keeping a tight feedback loop between the outcome a client actually needs and every technical decision made to get there.

Caleb Wharton: building AI products at RUSH

Caleb Wharton came at AI from a business and computer science background. He started a BCom majoring in accounting and in innovation and entrepreneurship, was exposed to coding through a first-year paper, and after an entrepreneurship programme that put him alongside computer science students, got hooked enough to pick up a computer science degree alongside his commerce one.

His first experience of ChatGPT in 2023 was a crazy moment, as it was for many, when he realised he could automate his homework, coding and essays, but he saw it as a tool rather than a career. Even the 2025 wave of agentic coding tools like Claude Code did not immediately change his mindset. The shift came after a couple of months overseas: back working on personal projects with GitHub Copilot, the jump in the quality and efficiency of the code the models produced was his lightbulb moment. He decided the software industry would not be the same, and that keeping a place in it meant changing how he thought about software and AI entirely.

He took himself from basic machine learning knowledge to a real understanding of large language models, agentic coding and computer vision, building AI-powered applications through rapid coding. That led him to RUSH, joining its newly founded AI chapter. RUSH is a digital services agency that delivers software end to end, from design and scoping through execution to managed services and IoT, under the line "we design and build technology to better serve humankind". Its work includes the MetService weather app, the country's COVID Tracer app with its 804 million QR scans, and the Z app with over 1.26 million coffees ordered a year.

His role sits across three pillars. The first is building AI tools and products for RUSH and its clients, from agentic workflows that take work off internal teams to enterprise applications and off-the-shelf models. The second is AI enablement, bringing both technical and non-technical people up to speed and helping them rework their workflows around AI. The third is AI-assisted rapid development, which the agency uses to iterate quickly through the discovery phase with clients and ship high-quality software faster.

Among his current projects: an organisational brain that gives each project a single repo-based source of truth in markdown alongside the codebase, plus discrete workflows codified into skills to speed up delivery and standardise quality; end-to-end automation of a large New Zealand enterprise's procurement process, which has been a useful lesson in where AI helps, where simple visualisation beats an agentic workflow, and where agentic workflows earn their keep; and an AI competency framework that another enterprise's departments and staff can assess themselves against to uplift their workforce.

His lessons are sharp. Sharpen your axe: borrowing Lincoln's line about spending four of six hours sharpening the axe, he argues you have to think critically about a system before you build it, because AI is great for execution but you cannot outsource your critical thinking, and the last 20% is always the hardest. Think for the future: an AI engineer's job is less about writing code and more about defining outcomes, working out the inputs and getting from point to point, so high-level systems thinking beats memorising the syntax of a language that may soon be irrelevant. And enthusiasm matters over experience, because the field changes week to week and no one can claim to know everything.

On rolling AI out in larger, more risk-averse organisations, his view is that the directive has to come from the top. The enterprises that succeed set clear policy and strategy for how AI should be used, then work down level by level with practical, role-relevant wins that turn sceptics into advocates.

His tips for getting in: be curious and stay immersed through podcasts and research papers; build, build, build, because getting on the tools beats learning passively; and talk about it more. However good you are, if no one knows who you are it is irrelevant, so put yourself out there at events, hackathons and online. As he puts it, things come to people who ask for them.