Hiring Data Scientists
JeanFrancoisPuget 2700028FGP Visits (13447)
Hiring data scientists can be one of the hardest job these days. One might think that it is hard because of lack of available candidates. I'd say it is just the opposite. Data science is so hot that many people rebrand themselves as data scientists. One of the issue when hiring is to discover among the many candidates which ones are true data scientists, and which ones pretend to be data scientists.
Here is how I proceed when I interview some applicants. It is by no mean an official IBM statement about what data scientists should be. It is not representative of what other IBMers might do. It may not even be useful to anyone but me. Here it is anyway. Read at your own risk!
I am not interviewing data scientists to perform scientific research or otherwise valuable activity. I am interviewing data scientists that will help our clients improve their businesses. This biases what I am looking for.
In a nutshell, I am looking for problem solvers.
It is no wonder why I have a good a priori when the candidate includes problem solving information in her resume, for instance participation in some Kagg
I won't disclose the kind of problems I use for obvious reasons, but I found the above picture to be representative of the interviews I run. Another great example is New
Of course, problem solving is not the only thing I look for. Background in some key areas of data science such as statistics, machine learning, operations research, or economics is mandatory. Background in related disciplines like signal processing or quantitative finance is good. Ability to use tools such as R or Python, or commercial software like SPSS or its competitors is almost mandatory. Software programming skills are definitely a plus, especially if one does not use the previous tools. Working knowledge of distributed approaches such as Spark is becoming more and more important. But all these are only useful if they can be put together to solve problems.