From classical piano to AI engineer - my interview with Rory Preddy at Devconf

Earlier this year I was interviewed live at Devconf by Rory Preddy, Principal Cloud Advocate at Microsoft. Microsoft shared it across their global social media channels, where it reached over 20 million people - which still feels a bit surreal.

The actual origin story

Classical piano came first. I used the word "compulsion" in the interview rather than "passion", which I think is more honest - it was never really a choice.

I started programming Java at school, liked it, finished matric and decided I wanted to be a concert pianist. My parents gently suggested I earn money first. So I enrolled in computer science at UP - and eight months in, deep in C++, I decided it was too dry and deregistered. Not a decision I'd necessarily recommend, but it led somewhere good: I spent that year focused entirely on piano lessons with a professor at the university, letting the compulsion run properly for once.

The following year I somehow ended up in a statistics and economics degree. It turns out statistics is a surprisingly colourful field. The question of whether an event is going to happen or not is a pretty interesting one to spend four years with. There's also something in the way both music and statistics train you to find structure inside noise - I think that connection is more than cosmetic, and it came up in the conversation with Rory.

How I ended up in AI engineering

I joined the industry as a data scientist and thought that was what I was. But the pull towards engineering kept showing up.

An early turning point was spending time containerising applications previously deployed to virtual machines and moving them onto Kubernetes - a great on-ramp into cloud. That led to deep Azure work, including a year in the UK, and eventually into AI.

The real entry point into AI engineering came when I spotted an opportunity to bring RAG into an internal LLM tool - letting it draw on organisation-specific knowledge rather than just what it was trained on. I thought it would be useful, so I built it. That was the start. From there came more AI applications, paved paths for others to follow and, eventually, chairing an AI guild - a weekly, completely informal, self-organised group where anyone curious about AI can pitch up, share a use case and pair with someone more experienced. No obligation, no leadership mandate. People just arrive. I love that.

The thread through all of it

Piano, statistics, data science, engineering, AI - it isn't a tidy narrative I had planned. It's curiosity followed consistently. None of my moves looked obviously sensible at the time: deregistering from a degree, studying a subject I'd never heard of, volunteering for projects outside my job description. But each one led somewhere.

Rory asked near the end what I'd say to people who feel like they don't fit a particular mould. The honest answer is that not fitting a mould can be an advantage - as long as you find somewhere that lets you be useful precisely because of it, rather than in spite of it.

You can find the original posts from Microsoft and Rory on LinkedIn.