I’ve been delaying writing a new post here for a while. In my defense, I’ve written thousands of lines of Python code (most of it open-source). On the other hand, I’ve been busy giving talks, lots of them.
If you checkout my speakerdeck, or the talks section, you’ll notice that they became a bit repetitive, as I was waiting 4 gtx1080ti GPUs to arrive for continuing the development of deep learning models for extra-solar planets direct detection. Basically, I had to spin the same ideas in different, hopefully interesting, ways. One presentation focused on the experience of building Python tools that are useful for other researchers (PySciDataGre group kick-off meeting), other on the idea of organizing a data challenge for high-contrast (Paris-Saclay Center for Data Science) and the rest of them on my experience doing data science in an academic context (Grenoble Alpes DataClub, Data Science in the Alps workshop, European Week of Astronomy and Space Science 2018).
At the European Week of Astronomy and Space Science, I found myself involved in a very interesting discussion around what Open Science is, reproducibility, and good practices for publishing data and code along with academic papers. This is a conflicting issue for me personally, as I’m developing two packages and only one of them is open-source (VIP). My view is that, for us young researchers, the overhead of publishing (as a package or workflow), every single code we develop is larger than the benefits it would bring in terms of our future (permanent) academic career. This hopefully will change in the future, as a new generation of researchers will outnumber the
old current generation, the obsolete nature of the scientific paper is redefined, and as the metrics for career advancement change to take into account the skills of the new breed of academic.
I’m writing this post during the first session of the Python in Astronomy 2018 meeting (yay!) at the Flatiron Institute in Manhattan, NYC. I’m very excited to be here and that definitively will be the topic of my next post.