And how are things in… Nijmegen?
It was bound to happen after my lamentation about an internship abroad: I’m writing a column about what’s it like doing an internship in the Netherlands, or to be more specific, in Nijmegen. Completely in accordance with my hypothesis, it turns out that an internship at home can be almost similar to an internship abroad.
The main difference is probably that I can’t organize a daytrip with colleagues to the top of an active volcano, or go on a hiking trip on a glacier – the limitations of the Netherlands. Instead, I commute on the train every day (a meager hour from door to door) – no great effort if you’re used to spending two hours on the train travelling from your parents to Eindhoven.
What on earth am I doing? The Radboud UMC has a large pathology department where pathologists spend a considerable amount of their time looking at (tissue)biotopes in order to make a diagnosis. Staring at small stained cells on a glass plate through a microscope or on an enormous computer screen is knowledge-intensive (read: years of training) and boring at the same time. The perfect task to automatize via the computer, via deep neural networks to be precise.
The thing I noticed most during my first weeks was how out of my depth I felt: my first time doing some serious programming using a stack of existing in-house-software, designing neutral networks, optimizing. It made me realize how most of the courses during my bachelor’s program didn’t prepare me substantively for this internship. Perhaps this is the curse of an interdisciplinary engineer’s bachelor’s program: it starts broad-based but can’t end with an in-depth study because you have to bundle the knowledge of several groups per department into a three-year package.
On the other hand, maybe everyone here suffers from the same problems at times, because only a small percentage has an actual background in computer science. I sit across the desk from my daily supervisor who has obtained a degree in medicine and who was in the middle of his specialization track to become a pathologist when he decided to ‘do something else.’ And the same applies to two biologists-turned-data-scientists sitting at other flex desks.
Deep learning, specifically when applied to medical data, is still in its infancy and it comes as no surprise that universities haven’t been able to completely adjust their programs to this abruptly rising field of research. By the end of this internship (only five more weeks to go) I will hopefully be able to say that I have become at least somewhat skilled in this field. What I can say in any case, is that I am enjoying myself and that I don’t regret not being able to go on daytrips in the area surrounding a university in a far-away country.