About

I started in AdTech in 2011 as the only front-end engineer on a platform rebuild, asked to move to the back end four years later, and got it — then spent a year and a half building a self-serve advertising marketplace, reporting to the CTO.

In 2017 I left to start my own company, mostly because I had spent six years sitting next to commercial people and realised I did not understand how they thought. For two years I did the whole arc: sold directly, ran the meetings, scoped the projects and built them. Insurance agent hierarchies, booking systems, tournament registration, and fifteen-odd automation pipelines for customers who needed a price list watched rather than a platform built.

Innity asked me back twice over those two years. I returned in 2019 at a higher grade and stayed through COVID, working on backends carrying tens of thousands of requests a second.

Then I co-founded an e-commerce venture, and closed it. The market had cooled, the product line was one none of us had any feel for, and we stopped while there was still money in the account rather than spend it proving a thesis that had already failed. I learned more from that than from anything that worked: hold no inventory you do not need to hold, and do not build what you can buy.

After that came a year inside a global logistics operator as a business process specialist, which is where my enterprise AI work actually began — hybrid OCR and LLM document processing, RPA across finance and logistics, and a hard education in integrating with systems that have no API and cannot be replaced. I then led client delivery at a software agency, and took six deliberate months to work on frontier AI full time, because the tooling shifts substantially every three to six months and I wanted to be current rather than adjacent.

For the last three years I have mostly worked on inherited systems — code somebody else wrote, usually after the person who owned it had already left. Reading an undocumented system quickly, finding where it actually fails rather than where it appears to, and stabilising it before rebuilding: that has become the thing I am good at.

That is also, more or less, the description of a forward deployed engineer. It is where I want the next chapter to be — inside a customer’s environment, holding the whole line from the first conversation to the thing running in production on a Monday morning.

I work in English, Mandarin and Bahasa Malaysia, which in this region is less a credential than a practical requirement.