weirdtunguska

Sure. I consider that the way we do data science and hire data scientists is a bubble because it´s unsusteinable and business are asking more and more about the returns of the large, expensive investments on "artificial intelligence". No doubt that working with data is getting more and more valuable, but the way it is done today is changing, and fast.

From a data scientist that has been working as an architect for data systems for some time now, the way that data science was done 2 years ago changed considerably to how we do data science and design data systems geared to machine learning today. Data can be the "fuel" the economy, but the extreme demand for data scientists that need to know a huge number of things, from statistics do machine learning to programming, databases and so on, is reducing, and more specialized roles, like machine learning engineer, is growing. That´s the reason for my comment: the bubble of data scientist highly skilled over a vast number of areas is dying, and data scientists need to start to specialize to survive.

As a physicist by training that has been working as a data scientist and believing that the data science bubble will burst sooner rather than later, are these certifications a good path to help change course and keep myself relevant?

Just to point out, in Brasil seatbelts are also required by law, so it´s not possible that "there are no seatbelts because it´s cheaper".

- FLAVOR: Ubuntu Desktop

- HEADLINE: Make KDE again a first class citizen

- DESCRIPTION: Kubuntu used to be very similar to the Ubuntu distribution and now, because of the "fork", it´s drifting. It is also very different on configuration, packages and behavior when doing an `apt-get install kde-desktop` on an Ubuntu installation versus Kubuntu, and it should be the same. - ROLE/AFFILIATION: Research Scientist on a large Multinational

- FLAVOR: Ubuntu Desktop, Ubuntu Server

- HEADLINE: Full compatibility with Debian packages and paths

- DESCRIPTION: Please, please keep package and paths compatibility with Debian. The amount of work to get Debian packages work on a recent Ubuntu distribution is huge, and there are a lot of scientific software that is geared to Debian, not Ubuntu.

- ROLE/AFFILIATION: Research Scientist on a large Multinational

Is it restricted to gnome?

Why can't this be done using standard ways, like deb packages?

I used Chebyshev approximation on my thesis connecting a boundary value problem to a way to solve it using Fourier Transform. Very, very useful when solving a system of equations that can diverge if a iterative solver is used.

Care to explain why for the non design folks?