I started in design, taught myself to build, and ended up running web engineering for Intuit across Asia Pacific. This is the short version of twenty years, written for anyone deciding whether to hire me.
I started as a designer and taught myself to code along the way. Then I spent seven years at Intuit. I joined as the one web developer for QuickBooks across about 200 countries. I left as the engineering manager delivering the company's largest launches in the region.
Somewhere in the middle I worked something out about myself. I come up with ideas much faster than I can build them. So I stopped trying to be the fastest developer in the room. I started building what lets more get done: first a team, then an agency partnership, and now AI agents.
Outside work I build machine learning systems. One searches the sky for nebulae nobody has found yet. The newest one looks for fires in satellite images of Earth.
Twelve years as a graphic, digital and UX designer. I worked in-house at Ingram Micro, Coverforce and Fairfax Media, ran a small studio of my own, and freelanced for small businesses. Web development started as a side skill in those jobs and slowly took over. Looking back, the pattern was already there: I could always picture more than I could make on my own.
I joined Intuit as the only web developer for QuickBooks' emerging markets. That was roughly 200 countries on one shared platform, with eight regional sites (later thirteen). It eventually covered eleven languages and six currencies. Most regional teams looked after one country. Mine looked after everyone else.
Over five years the international business grew about 4× in revenue and 3× in customers, and went from losing money to being strongly profitable. That was a whole team's work, not mine alone. But the website I ran was its cheapest way to win customers, by about ten times. From 2021 I owned its forecast myself (a number that usually sits with a marketer, not an engineer).
I also took the region from English only to eleven languages. I tested the idea with a small Spanish sign-up flow first, and once it worked I built the process that rolled out nine more. Along the way I presented the personalised web experience I'd built, in person, to Intuit co-founder Scott Cook and to Alex Chriss, now CEO of PayPal.
I was promoted to Staff and did some maths. Me building on my own was worth less than me directing ten people who build. So I found the team's first agency partner and grew it from a $30k engagement into an embedded team of eight. That setup is still running four years later.
I was managing two engineers and carrying commercial numbers while still on an individual contributor title. An internal re-levelling later rated the role at Principal level on most competencies. Over the same stretch I doubled the conversion rate of the global pricing page.
I became Engineering Manager for APAC web, covering QuickBooks and Mailchimp. The region was 65 countries and 14 priority markets, with more than 70 million visitors a year and about A$0.5M in budget. I led the regional delivery of Intuit's biggest brand overhaul in its modern history. The launch shipped over 40,000 lines of code with no outages that blocked a customer.
When Mailchimp came in as a second brand, we took it on without extra headcount. Instead of hiring, I changed how work reached the team. I was also the first leader in the division to pilot AI for handling incoming work requests. The UK and Canada teams later built their own versions from my framework.
While the day job grew, I kept building my own things at night. There's a deep learning system that searches nine-band survey images for undiscovered nebulae, and an automated pipeline for my observatory. There's an open-source planning tool with a plugin in the official marketplace for N.I.N.A. (the imaging software a lot of amateur astronomers run). And there's Tellisto, a client gallery platform for photographers.
These are built properly, with tests, CI, monitoring and real users. I'm also president of the Northern Sydney Astronomical Society, where I end up building the club's software too.
Intuit made my role redundant in its 2026 restructure, and I'm using the time on purpose. I've taken the nebula detection model and adapted it to find active fires in satellite images, and I'm looking for the next team to lead.
AI has finally closed the gap between how fast I think and how fast I can build. I'd like to bring that to a team working on something that matters.
“Calm and craft… a testament to your leadership.”VP, APAC · on the brand overhaul launch
“Great work on the release… excellent collaboration.”SVP, International
“You've made a major difference for us.”UK web lead · after I fixed a billing outage that wasn't mine to fix
“Rohan is a superstar, especially when you consider the number of markets and teams he supports every day.”Peer feedback · from my Staff promotion case
International Impact Through Collaboration award (2025) · Stronger Together award (2022) · Learning Leader, top-5 learner internationally (2020)
If you're the person I report to, give me the business goal and the number, and I'll run the function so you don't have to. I carry the forecast and the budget myself, and I report like I own the result instead of sending status updates.
With my team, engineers get autonomy and I hold us to outcomes. They own how it gets built. I own what we build, why, and the risks. When someone's stuck or slipping I try to find out why early. Usually they were briefed badly, they're overloaded, or they're blocked on another team, so I fix the cause. I also take on the meeting load and the org noise so senior engineers can do senior work. Doing that freed up 30–40% of my team's capacity at Intuit.
With the people above me, I don't do surprises. I raise risks while they're still cheap to fix, and an escalation from me is a short summary with my recommendation. I disagree in private and back the decision in public, and credit goes to whoever did the work. My team stayed together through a restructure, and I think that's part of the reason.
A U-Net++ segmentation model that searches 9-band survey images for undiscovered nebulae. It trains on 13,000 tiles matched up from more than 12 sky surveys, with experiment tracking, data versioning and CI. It has already found real candidates.
The same architecture pointed at Earth. It finds fires on the standard Landsat-8 benchmark, and I'm testing different encoders against each other on one shared evaluation setup. Still in progress.
A client gallery and CRM platform for photographers. Each studio's data is kept separate at the database level, and uploads are scanned for unsafe content. Billing runs through Stripe. It has 600+ automated tests, and it's now open to everyone.
Turns a multi-terabyte image archive into a sky map you can search, with a companion plugin in N.I.N.A.'s official marketplace. Free and MIT-licensed, built for the community.
My list of ideas has always been longer than my list of finished things. That gap is why I care so much about teams and tools that let me build more.