Juliet is a principal software engineer at UNSILO.
I’m originally from Johannesburg, but I’ve lived in Denmark for more than 20 years. I qualified as a physicist and I was actually working in quantum field theory, which I studied for my master’s degree in South Africa. As part of that course I did some programming in Fortran. When I came to Denmark, there were very few jobs in physics (especially for a specialist in quantum physics) and I took a full-time developer course in Java. I worked for five years with DSG, a large supermarket chain in Denmark, where I was working on mainframes. After a six-month period being outsourced to IBM, I moved to work in healthcare with one of CSC’s Danish offices. Here we were working with regional government and dealing with clinical data and discharge summaries. These were very big projects over long periods – one of the last projects I worked on comprised over 3000 hours of developer time.
Can you describe a typical day at work here at UNSILO?
Well, I’ve only been here a few weeks, but we are working on one of UNSILO’s software modules for publishers. In the last few years my work has largely been around integration between systems, and currently my team working on transforming the formats in which the publishers deliver material to something we can use. There are four of us in our team, and we are working on the results generated by the machine-learning engine and how they are presented to the user.
How do you find working in machine learning?
I’m not doing the machine learning directly, but of course you get to know about the application. It’s interesting because it is more imprecise, and a bit of a challenge. At UNSILO, you are working with fuzzy logic and probabilities.
How do you find working with a startup?
After several years with a big company in healthcare, it’s a bit of a culture shock – things move fast here! Everything is more agile, in all senses of the word. There is a willingness to do things differently that is very refreshing. And there are no servers in the office!
What would you be doing if you weren’t a developer?
I enjoy doing what I do. I suppose if I had continued as a physicist I would still be working in quantum field theory; but there is quite a bit of overlap between physics and what I am doing today. Machine learning is all about uncertainty, and of course there are parallels in physics, with Heisenberg’s Uncertainty Principle. There are other areas in physics as well that deal in uncertainty, for example thermal and statistical physics.
I like the modelling part of software development, which you also do in physics. You make use of modelling in the object-oriented software process by introducing concepts. You identify a concept model, and this model is documented as part of the development process. The models have to be as simple as possible otherwise others don’t follow the modelling (and that misses the point). It helps if you are good at logically grouping things. I always feel a good piece of code makes good sense; it is satisfying. In that way, my physics training helps here – I have learned to work with abstract things and turn them into something concrete, concentrating on just the most important details.
So you speak Danish? Yes, I have two teenage kids, aged 16 and 18, and we speak Danish in the family at home (although I still have a bit of a foreign accent!)
Many thanks, Juliet.