How to get a Data Science job?
Let’s apply Data Science to your business.
--
More than ten years ago (when many “big data” methods were just beginning to emerge), data scientists were seen as a high risk, high reward, highly prestigious profession. It was rare that anyone would be promoted by publishing a single paper, but when they did, it could be used to enhance reputation and later on, tenure.
And the beginning of “big data” was hardly mind blowing. It was kind of like putting a pile of rocks in a stream. If you have the rock, you have a data science process to replicate, process, and analyse the original, reanalyze the data, and voila, you have a new large data set. If not, well, perhaps you could convince yourself that it was still worth doing.
This didn’t change until 1990s
During the 1990s, this became the dominant paradigm for data scientists: Powerful machine learning platforms gave rise to more general applications, such as predictive analytics, cluster computing, storage management, research-based communication, and decision support software.
This led to the publication of the seminal book Data Science and the creation of the term “data scientist”, as a specific subfield.
For good reasons, developers don’t like to update their operating system. So once you have a map, it’s straightforward to stream these updates to clients.
So why don’t more companies hire data scientists?
There are a lot of reasons, and that’s for a blog post that will be written soon. But from my perspective as an analyst, I’ve been monitoring this space very closely, and there are three things we can call out.
Software infrastructure just made it cheaper and easier for any data scientist in the world to start developing data tools. Anyone anywhere can start.
To grow a project to a level where it’s worth paying someone else to help you build it, you need to do it fast. You need someone to absorb the code you write, produce stable, reproducible and usable code, and ship it to a commercial environment.
To make that happen, your web developers need reliable, agile frameworks, good tooling and continuous integration. If you can pay a developer to…
