Concluding his presentation with a little manifesto, Tomasz states that Data Science is not a technical discipline - it incorporates technical methods but, first and foremost, it is a conceptual discipline. The focus is on understanding the point of improving a process, usually a business one. It's not about using the tools, as the idea matters - the ability how to approach a specific issue.
The ultimate take away from the presentation is that a real business case has usually nothing to do with idealized world of models - it requires non standard approach. The fact is that person knowing just technical aspects of data science without the ability to think in a conceptual way usually is on a lost position - the world isn't like the Iris data set. Having said that Tomasz adds that in order to deliver non standard approaches one must understand every step in the algorithm. Black boxes will doom you to fail sooner or later.