According to a report by CyberCoders, the digital universe will reach 40 zettabytes (45 trillion gigabytes) by the end of the decade, a 50-fold growth. This brings us all to one question, “What do we do with such vast amounts of data?”
As Jonathan Shaw pointed out in Harvard Magazine, it’s not the amount of data that makes it a really big deal, it’s the ability to actually do something with it.This trend leads to the emergence of a new set of folks - known as data scientists who help in extracting meaning from the data and help businesses take important decisions based on that.
So, why data scientists are so difficult to hire?
Data Science is an extremely complex domain, where the candidates are expected to have fairly strong programming skills and statistical understanding of ML. According to this report, data scientists are one of the most in-demand jobs in the world. Given the high requirements for this position, there is a huge talent gap — McKinsey Global Institute estimates the shortage of data scientists in 2018 at 190,000. And the number takes a bigger shape when it comes to senior data scientists.
So, in this circumstance, when everyone is competing for the very limited amount of talent, how can you make sure that you land up building a great data science team?
Know what to look for
Typically, a company should look at candidates with fairly strong programming skills and statistical understanding of ML. Exposure to big data platforms like Hadoop and Spark would be added bonus. Decision making, random forest are other topics generally vetted in a data science interview. You should see if the candidates have an understanding of what goes on in these and the knowledge of which problem should not be put into these.
More than implementing the algorithm what matters more is, given some basic input of data that's available, if they are able to make the choice of selecting the proper tools and algorithm.
As a business, you should be very clear about your requirements and the core competencies required to ensure data scientist are able to develop models that support and benefit enterprises.
And the most important part is a good JD. Create a proper JD focusing on your business needs, stating all the must-have and the good to have skills.
Broaden your target
When the demand is more than the supply, you can’t be very stringent with your requirements. Data Science is a very specialized skill and the no of people with 5+ years of experience are difficult to find in this domain. We have often seen companies mention a dozen of criteria who fail to land up with one single hire because they are being too stringent with each and every skill.
According to Deloitte, leading IT consultancy, most universities and colleges aren’t able to produce data scientists fast enough to keep pace with industry’s demands. There are not sufficiently trained data science professionals, most of them are trained on the job fresh out of their graduation. Some would come from engineering, mathematics or statistics background and some of them would be trained some institutes like Aegis School of Data Science. There are various courses available online as well, but you cannot expect a great professional from a course of 6 months or a year.
So, you need to structure your team accordingly. You can get freshers and train them in-house, of course take any decision after a proper cost-benefit analysis.
Ask yourself - Do you need a full-time guy?
There is a different alternative to getting your projects done as opposed to the traditional ways of headhunting. Websites like Kaggle, Analytics Vidhya, conduct contests and offer huge prize money for those who rank well in those contests. These can be as good as a job experience. How these contests come out is, suppose you are a company and want to build a recommendation system for yourself. You can recruit say 4 data scientists who can build that for you which is great, but on the hindsight, let’s face it, data scientists do not come in cheap. What if the recommendation system does not work after you pay them a salary for one year. There is a huge opportunity cost involved.
Instead of that, you put up a contest of 1 million dollars on Kaggle and pay that amount only to the guy who can provide you with a satisfactory solution. Since it's a global contest, you also get to choose from the best of the best. But of course, if your company is heavily based on data, you would need an in-house person as well. But you can always outsource some tasks.
For startups - when to hire?
As we all know, startups run on a tight budget and it becomes difficult to have an experienced data science person in the initial stages. Again, it totally depends on the business. For example, in an e-commerce system, where 70% of your sales is given to the recommendation system, a data scientist is probably the key element in the organization. Ideally, it’s best to be in touch with someone when you are initially planning the startup.
Even if you are not going for a full-time hire, rope him in and he is going to advise and guide you, perhaps as a consultant during the early stages (given he really believes in what you’re doing). Today, the world is data-driven; we lag behind in keeping track of these data, and we end by saying that the world is chaotic instead of following and analyzing those data sources. Thus, even if you’re a startup the early you start planning, the better it is.
Don’t ignore business-sense
A right candidate does not only know his job well but someone who is well-aware of the user journey or user behavior from a business perspective. That’s why you would often see management graduates moving in the field of data science and data analytics as well. Always choose someone who understands your business and your user.
So, these are the 5 things startups should do to get great data scientists on board. Did we miss any? Let us know in the comments.