Kelvin Kaifung Chan

I'm co-founder and CTO at IRL, we build toys, tools, and environments that help people learn through play.

Currently focused on building IRL PLAYground.

Some things about me:

  • Grew up in Hong Kong
  • Trained as an architect at the Architectural Association
  • Previously designed schools and learning spaces at a non-profit architecture studio
  • Based in London

Questions

Some questions I find interesting at the moment, aside from those related directly to IRL and IRL PLAYground.

How does a self-taught approach shape technical decisions and engineering culture?

Without formal training, engineers often develop judgment through building, learning to prioritise, adapt, and ask better questions. This can lead to faster iteration and stronger product intuition, but it may also introduce risks: unclear abstractions, inconsistent standards, and cultural gaps around documentation, review, and long-term maintenance. What values emerge when intuition, not curriculum, guides the work?

What makes a technical system legible not just to machines, but to teams?

It’s possible to write code that runs perfectly but leaves future collaborators guessing. As teams grow, legibility becomes not just a technical concern but a cultural one: documentation, naming, modularity, and even pacing all shape how a system can be understood and changed. What does it mean to build in a way that stays readable not just at runtime, but over time?

What role does ambiguity play in technical work?

Engineering often pushes toward clarity, precision, and closure. But much of product development, especially early on, requires holding multiple possibilities open. Ambiguity can be productive, even necessary. Where is it helpful, and where does it become a liability?

Are the mindsets required for finding product–market fit and scaling fundamentally opposed?

Product–market fit demands exploration: speed, openness, and a willingness to work with ambiguity. Scaling requires optimisation: structure, discipline, and a bias toward repeatability. These aren’t just different stages, they often rely on different ways of thinking. What does it mean to bring these logics together? Are compromises inevitable, or can both mindsets coexist without one diluting the other?