Description
Data Scientists (or researchers) often do many early-stage prototyping and run many experiments before finding the optimal solution for the problem they are trying to solve. However, sometimes, a lot of the knowledge gained in the process stays with the person. In this proposed talk, I would like to introduce how to produce, share and ensure the reproducibility of experiments using jupyter notebooks and github. This is a series of lessons Iearned while being a research student (where open-sourcing the code and data used to generate papers is very important) and more recently, working in a new team in Zalando, where we have experimented with different ways to share and review the data science tasks, as well as the knowledge generated from those tasks.