Listen to the interview
You came from AI research at Imperial. What’s different about building AI that clients depend on every day?
The core of the work hasn’t changed, but the shape of it has. In academic research, the key thing is the idea; after that, you might wait years for a use case.
Here, the idea also needs execution. It’s a faster feedback loop, which can also turn into a never-ending improvement loop. It’s just as complex to navigate, but in a very different way.
You were on a hands-on client project on Day 1. What did that teach you, that coding alone wouldn’t?
Few places would put a new joiner in front of a client that early, and I really valued that trust. It also, however, comes with a higher level of responsibility and requires you to build up judgement that coding alone won’t teach you.
I was tasked to help improve our document extraction capabilities to be truly market-leading, and to apply that at the same time to live clients piloting our DDQ solution. This has meant going beyond research and exercising judgement about what’s worth building for the end use-case.
What kind of engineering profiles do you think fit best at Vega?
You need to have curiosity beyond what you’re being asked to do: the best engineers I have seen at Vega come up with really creative solutions because they have thought that much deeper about the problem itself.
You also need to have tolerance for problems without clear edges. We are given freedom here to move quickly, which is rewarding, and that also means constantly navigating new things and being okay with going outside of your comfort zone.