The Systems No One Saw Coming

In 2022, I built my first serious system: an AI-based search engine concept focused on intent rather than keyword matching. It was early and experimental, not polished, not production-ready. But it mattered enormously because it was the first time I wasn't just learning anymore, I was building systems that had direction and purpose. I wasn't following a tutorial or recreating something that already existed. I had an idea about how search could work differently, understanding what a user meant rather than just matching what they typed, and I built it from scratch. I was seventeen years old.

Let me be very clear about something, because it's important and because the world has a way of erasing who did what first when the "who" isn't from the right place: I built this before AI search engines and agentic systems became mainstream conversations. Before the hype. Before the venture capital flooding into anything with "AI" in the pitch deck. Before every startup on Earth suddenly had an "AI-powered semantic search" feature. At the time, it was not part of any trend. It was not a bandwagon I was jumping on. It was just an idea I couldn't ignore, an intuition that search should understand intent, not just keywords, and that the technology existed to make that possible, and that someone should build it. So I did. At seventeen. In Kenya. While the rest of the world was still arguing about whether remote work was productive.

I tested the concept. I built a working prototype. And the tech industry, years later, would "discover" the same idea and call it innovation.

Funny how that works. Funny how innovation has a geography. Funny how the same idea is "experimental" when it comes from Nairobi and "revolutionary" when it comes from San Francisco. That same year marked something that has not stopped since: I started applying for opportunities. Internships. Programs. Funding. Access to environments where I could build more seriously, with better tools, alongside other builders, with resources that didn't require me to choose between server time and food. And from 2022 all the way to now, that cycle has continued without interruption. Apply. Get rejected. Build more. Apply again. The rejections blur together into a background hum, like static on a radio you can't turn off. Each one stings less than the last, not because it hurts less, but because you develop scar tissue. You stop expecting yes. You start expecting the form letter, the generic rejection, the "we were impressed but" that always ends the same way.

But the building never stopped. That's the thing about being driven by something internal rather than external validation, the rejections hurt, they really do, but they don't stop the work. They can't. The work is the only thing that's mine.

Alongside all this, I started freelance software engineering. Small gigs, sometimes paid, sometimes not. The boundaries were often unclear, what was I building, for whom, under what terms. I built backend systems. Fixed issues in production environments. Contributed to projects where I was essentially an engineer without the title, the pay, or the stability that title is supposed to provide. I learned the term for this later: shadow engineering. Building real systems, doing real work, but without proportional recognition or stability. Visible enough to be used. Invisible enough to be underpaid. It's a category that exists everywhere but is rarely named, and African tech workers occupy it disproportionately. You are essential to the system but external to its rewards. You build what others take credit for. You fix what others broke, and your payment arrives late, partial, or not at all.

One of the early things I built during this time was a plagiarism detection tool. When I released it on GitHub, it reached seventeen stars. Seventeen. I know how that sounds. Objectively, seventeen GitHub stars is nothing. It's a rounding error on the internet. It's the number of people who accidentally clicked while scrolling. But for me, at that moment, it mattered enormously. It was the first time something I built had external recognition, proof that a stranger, somewhere in the world, found my work useful. Seventeen people I had never met, whose faces I will never see, looked at what I made and thought: this has value. I can use this.

When you build in isolation, when no one is watching, when you're not part of any community or ecosystem or network, you start to doubt whether what you're doing is real. Whether the systems in your head exist in anyone else's world. Whether you're actually an engineer or just someone playing at being one. Seventeen stars told me: you're real. Keep going. The number was small, but the signal was clear.