About

I have been asking “why?” for as long as I can remember. I asked it so often that my grandfather started calling me Why.

These days the question just has more places to go: computer science, mathematics, physics, astrophysics, machine learning, distributed systems, hardware, psychology, markets, cameras, storytelling. The pattern underneath is the same every time.

1. understand how it works→
2. take it apart→
3. build my own version→
4. ask if it could work differently

I would rather build a thing than only use it. Using something tells me what it does. Building it tells me why it was made that way, and whether it had to be.

I have been lucky to study and work around a lot of powerful ideas: pure and applied mathematics, physics, machine learning, computer vision, distributed and high performance computing, cloud infrastructure, large production systems. Most of them live inside universities, laboratories and big companies.

So the question I keep carrying home is what these ideas mean in Soshanguve.

• What does distributed computing mean to someone in a township who wants to start a farm but does not have the capital?
• What is a directed acyclic graph when the nodes are people, suppliers, taxis, money, land, skills and information?
• What does optimisation mean when someone only has R500?
• What does AI mean when your main computer is a phone, and data costs real money?
• How much of what a bank or a telecoms company can do could a small community rebuild with open source and cheap compute?

I have not answered these. They are not rhetorical. They are the questions that drive most of my work.

Knowledge, technology, networks and opportunity are not spread evenly, and most of the gap is not talent. It is who had internet early. Who knew the words. Who knew the right person, went to the right school, could afford the software, or happened to be near the opportunity.

Technology will not fix inequality on its own, and I do not believe anyone who says it will. My interest is narrower and more practical: can technology give more people a comparable starting point?

Sometimes that looks like education that opens on a phone, free learning material, an AI tutor, open source software, a website for a small township business, or a system that runs on ordinary hardware. Sometimes it is just an explanation written for someone who was never handed the background.

This is one of my most important tests. If I cannot explain something to my mother, my father, my grandfather, or anyone outside my field without breaking the idea, I probably do not understand it well enough yet. That is not about dumbing things down. It is translation.

The second half of the test matters more: can I help them use it? I care about the moment someone goes from “I don't understand this” to “I can actually do this myself.”

That is why I teach: talks, tutorials, workshops, practical sessions, online conversations, projects, informal lessons with family and friends, and collaborations with academics and other practitioners. Some of it is free, on purpose. Not as charity. I just enjoy sharing what I learn, and not every exchange of knowledge needs to become a transaction.

I do not want to be the person who builds everything for everybody forever. I would rather help someone understand enough to build the next thing without me.

My day job has shown me how large organisations solve problems: big datasets, cloud infrastructure, specialised teams, production platforms, governance, vendors, distributed systems moving millions of events, and real money behind all of it. I value that education.

It also left me with a question I cannot put down: what does the same capability look like when almost none of those resources are there? That is a big part of why open source pulls at me. Learn from the well resourced systems, then ask which principles survive the move to something smaller, cheaper and easier to reach.

Professionally I am a Lead Data Scientist, but the work has kept pulling me deeper, into engineering, infrastructure, architecture and systems. The lesson that did it:

A model sitting in a notebook is not impact.

Anyone can train a model. The interesting problem is everything after. Can people reach it? Can it scale? Does it survive failure, real data, and someone else maintaining it? Does it actually improve a decision?

1. idea→
2. model→
3. system→
4. user→
5. impact

I love modern AI. I build with agents and intelligent systems every week. But I keep wondering whether the current way of framing intelligence is the only useful one, or just the one we happened to arrive at.

People coordinated, survived, passed on knowledge and managed uncertainty for generations before computers existed. I am curious whether African collective philosophies, ubuntu and ubunye especially, have anything to say about how intelligent systems could work. Could agents coordinate differently? Are there useful models of intelligence we ignored? What would a system look like if some of its design principles came from communities, not only from corporations and computer architecture?

I want to be careful here. This is a question I am exploring, not a theory I have developed or a claim I can defend. I do not know the answer. That is why I find it interesting.

My projects are laboratories for these questions. They are not a portfolio of startups.

I don't want a machine that agrees with me faster. I want one that helps me discover when I'm wrong faster.

Lately I am learning hardware, and it is the same curiosity going one layer down. Software eventually reaches physics. The cloud is someone's machines. AI depends on chips. Memory, networks, energy, latency and heat all have limits, and cameras depend on sensors and glass.

So I am learning GPUs, local compute, clusters, networking, electronics, sensors and embedded systems. Mostly because I dislike the point where my understanding stops at “someone else handles that.”

Photography is not a side note for me. Cameras, old cameras, lenses, drones, lighting, film, street photography, architecture, landscapes, people, documentary work, music production, and travelling to document places. Cameras are where physics, optics, electronics, computation, art and memory all meet in one object.

A camera is an extremely technical machine built to do something deeply human: remember this.

Storytelling matters to me more every year. Technology can create capability, but stories are how ideas travel.

More of my photographs on Instagram

I love Naruto, Dragon Ball Z and One Punch Man. Part of the appeal is getting to live inside someone else's model of the world for a while: what drives them, what they are afraid of, what they do when they lose.

That is the same thing I find fascinating about psychology. Why do people think the way they do? Why do groups behave differently from the people in them? How do incentives change behaviour? How can two people live through the same thing and come away with completely different stories? Also, sometimes a man just punches things very hard, and that is enough.

I like to share, teach, build, and work alongside people, especially people who know things I do not. I would rather build a team, or a system, where people can run on their own than hover over anyone. I do not micromanage.

Meetings are not my favourite part of any job. I show up, I contribute, and I collaborate a lot. I just prefer useful work and useful conversations over process for its own sake.

Not everything I do fits the story above, and I am fine with that. Not every project needs a business case. Not every interest has to become a startup or make money. I read about astrophysics, dynamical systems, optics, strange algorithms, music production, and whatever catches my attention next, simply because I want to know.

I am also learning to rest. Sleeping. Travelling. Taking photographs. Watching anime. Spending time with people. Occasionally doing nothing at all. It turns out constant productivity is not a requirement. I am still getting used to that.

Understand how something works. Take it apart. Build my own version. Ask whether it could work differently. Then share it, so the next person has less distance to cover.

I am still figuring most of it out. That is the fun part.