Abstract: As data scientists, and engineers more broadly, we often find that we could offload significant amounts of our work to computers. Instead of building complex algorithms, we could learn relationships using machine learning. Instead of manually searching for optimal parameters for our algorithms, we could run a black box optimizer. Instead of spending a lot of time on optimizing our code we could just rent a bigger cloud instance and run it there. This talk explores different methods to reduce human work using large amounts of compute power. It shows how to extend existing tools such as jupyter notebooks and highlights some useful tools many data scientists might not be familiar with. It explores when it is a good idea to leave work to the computer and how to spot opportunities for applying the methods shown.
Bio: Jannes Klaas is a quantitative researcher with a background in economics and finance. Currently a graduate student at Oxford University, he previously led two machine learning bootcamps and worked with several financial companies on data driven applications and trading strategies. His active research interests include systemic risk as well as large-scale automated knowledge discovery.