Nobody has ever hired me for just one thing.

Learning to build, and learning why it paid to.

I studied computer science at Capital University of Science and Technology, and I was the kind of student who kept building things nobody assigned — an mp3 player in Swift, a management system for a local medical store, an Android game about kicking rival politicians out of parliament. Some were good. Most were not. All of them taught me more than the coursework did.

I also started taking freelance work while still studying. Partly curiosity, but honestly, mostly arithmetic: what a developer earns locally and what the same work earns abroad are different numbers, and I wanted the second one. That decision shaped the next decade more than any single technology I learned.

Four years where bugs grounded aircraft.

I joined a UAV dependability lab straight out of university and stayed four years. The work was unlike anything a web job teaches you: ground control software for drones flying in swarm and formation, onboard autonomy running on companion computers so an aircraft keeps working when the radio link drops, and analysis tooling for wide-band signal data.

The part that changed me was not the code. It was standing in a field watching something I had written fly. I trained to fly the aircraft myself so I could test properly, because simulation tells you what you modelled and flight tells you what you missed. When your software has a physical consequence, you stop treating "it worked on my machine" as an answer.

The other half of the work was signals: analysis and direction-finding software running on instrumentation hardware, deciding which spikes in a noisy spectrum were real transmitters and where they were.

I started as a software engineer and left as Team Lead, running a team of six, hiring for the lab, and mentoring five interns through its programme.

Going independent, and learning the other half of engineering.

My first serious contract — a discovery platform for a Nevada company — overlapped my last months at the lab. It went well enough that I stopped choosing between the two and left in January 2023 to contract full time.

What followed taught me the half of software engineering that research work never does: scoping honestly, quoting a number and standing behind it, explaining a technical trade-off to someone who does not care about the technology, and shipping when a client's revenue depends on the date. I delivered more than seventy-five websites and platforms for clients across the US, UK and Europe, earned Legend status — the top tier of the Wix Partner Program — and was vetted into Fiverr Pro.

The strongest thing I took from it is the client feedback further down this page. Repeat work and referrals are a harder metric to game than a rating.

From developer to CTO, in three and a half years.

I joined YouGuide as a software developer on a contract basis. Three and a half years later I am its Chief Technology Officer, having moved through Head of Development on the way. The company is based in Diegem, Belgium, and I have worked with them remotely from Pakistan the entire time. It is small — around ten people, with two engineers reporting to me — and I want to be straight about that, because the title matters less than what it was earned for.

What it was earned for is the content engine. I built the AI generation and publishing pipeline that has produced over fifty thousand travel and language guides across multiple languages. The hard problem was never generating text; it was trust. Guides confidently describing places that do not exist are worse than no guides at all, so I built a validation layer that verifies every referenced location is real before anything publishes, and tightened the prompting until the failure rate fell.

That pipeline now earns its keep. The guides are licensed through partnerships with several of the world's leading AI companies, and licensing has become a significant revenue line for the business. I generalised it into a reusable platform so two sibling products — a technical eBook catalogue and a botanical guide service — could run on the same core without a rewrite, and built out an eSIM retail channel on Amazon alongside it.

Still building.

Seven years across domains that aren't supposed to overlap — drones and web platforms, research labs and revenue targets, writing the code and owning the architecture. All of it remotely, from Pakistan, for teams in Europe and North America.

What interests me now is what interested me at the start: problems where the answer isn't obvious yet, and where somebody has to be accountable for it actually working. If that sounds like something you're building, I'd like to hear about it.