What sparked your interest in physics?

Of all the subjects that were available to study at school, physics was the one that most appealed to my experimental instincts. I’m not a spectacularly good mathematician. I’m more of an engineer, and I like mucking about with stuff – I like building things and designing things and seeing what happens when you try them out. You get some of that in design and technology courses, too, but school-level physics is very experimental, because most of the mathematical bits end up being taught in the mathematics course as applied maths. So when I went to the University of Cambridge in the 1990s, I started out in physics even though, looking back, I think I had more of a natural affinity for engineering.

In my first year at Cambridge, the course was around 25% computer science, 25% physics, 25% materials science and 25% maths for physics. I enjoyed the experimental parts of the course, and especially the weekly sessions at the Cavendish lab. But the second year was two lots of physics, plus maths for physics, and after that it became clear that I was an engineer more than I was a physicist, so I moved across to engineering.

Why did you decide to go into computer science for your postgraduate work?

I’d been a computer programmer since I was a kid and, on some level, the Raspberry Pi is an attempt to recreate the positive aspects of how people like me learned computing back in the 1980s. I had a BBC Micro computer at school and at home, and a Commodore Amiga at home as well, so I had access to all these programmable machines starting from when I was about 10.

In my postgraduate work, I drifted into working purely on software, designing compilers and programming tools, but I probably went too far in the abstract direction. The place where I’ve ended up is closer to silicon engineering or electrical engineering. The former is kind of a software job these days, now that human beings aren’t drawing polygons that turn into bits of masks on silicon chips anymore. Instead, they’re writing descriptions of the chip’s behaviour in high-level languages and leaving the rest up to the tools they’ve developed. But there’s also an aspect of hands-on work in what I do – the actual grungy bit of getting a PCB [printed circuit board] and stapling stuff down on it to make a physical product you can sell. After a period of oscillation, I guess I ended up somewhere that’s right for me.

How (if at all) has your training in physics helped you in your career?

The maths has been very useful, and the analytical skills have been useful as well for the engineering end of things – basic experimental design, for example. We built the Raspberry Pi 4 recently, and after we put it out “into the wild” we got a lot of data coming in from users on their experiences, saying that this thing is surprisingly fast, or this other feature is surprisingly slow. We like to think that we design our products analytically, but even a small and relatively simple computer like the Raspberry Pi is complicated enough that after you build it, it becomes its own little system, with its own physics. To understand its performance, you need to design experiments to probe its behaviour, and that’s something I took away from my experience with physics.

• Read more about Eben Upton’s work and applications of small computers to the industrial “Internet of Things” in this month’s Physics World Focus on Computing