According to the World Economic Forum, data scientists will be the most sought-after profession by 2020. Their average salaries are among the highest in the world. For the USA alone, demand has increased six-fold in the past five years and data scientists will continue to be the most actively sought professional profile in the coming five years. IBM, for example, estimates that recruitments of data scientists will reach 700,000 by 2020.

Is this speciality going to be "the sexiest job of the 21st century"? In any event, it calls for exceptional skills, as Hal Varian, chief economist at Google, points out: a combination of a scientific background and analytical and IT skills, where business encounters IT in a very competitive market subject to constant evolution.

Accordingly, skilled recruitment and the ability to retain these strategically vital professionals prove to be essential for any company. In the IA business, where competition is fierce and innovation extremely fast-moving, an error in the choice of data scientists can impede or even halt a company's development, whereas the presence of the right experts will ratchet up its growth significantly.

Alas, recruitment is far from being an exact science.

So what determines the makings of a good data scientist?

There is an abundance of questionnaires for interviewing data scientists on the Internet. They are so widespread that their questions do not allow a candidate with the right profile to be selected. Rather than recruiting a candidate skilled at answering set questions, let's have a look at the essential qualities required of a good recruit.

 1. Data science, data science, data science

 Caution: open door. You will need to assess the candidates' technical knowledge. Do they master mathematical and statistical concepts? Are they exceptional programmers with black belts in hacking? You expect that, and we'll not dwell on it.

Data Science Venn Diagram 071808

2. Business sense

 Even if your data scientists can do R&D, they are not just theoreticians. They are at the heart of your business model. They have to understand your company and the constraints of your market, as well as sharing your ambitions.

 3. Communication skills

Good data scientists must be capable of interpreting their algorithms and communicating them to interested players in the company. This means contextualising a problem and interpreting it and its solution to colleagues from very different backgrounds. This includes written communication with a summary or report, oral communication which can proceed by iteration, i.e. by successive approximations, and visual communication which will simplify the data using diagrams, graphs or wireframes.

4. Project management, mental agility and a gift for teamwork

 The ability to work in a fast-moving process is essential. They must be able to work in a team whenever necessary with software engineers, functional experts, product managers, marketing specialists and designers.

5. Curiosity, creativity, reactivity

 Good data scientists must be able to see beyond established hypotheses and models and know how to solve problems creatively while always keeping abreast of the latest trends. They must be capable of sifting through both unknown formats and existing codes. Consequently, technical flexibility is as essential as experience, because in data analysis, standards are replaced and progress takes place at breakneck pace.

So recruiting good data scientists means, above all, being able to recognise the professional profile best suited to a company's DNA, its values and its projects. This is followed by spotting not only their scientific and technical skills, but also their abilities to communicate their results in intelligible, multidisciplinary language. It also means recognising their ability to work in a team to create a finished product. And finally, it means identifying in them a curious and agile mind with enough reactivity and creativity to get to grips rapidly with new data and adapt to every situation.

Originally published on August 17, 2018 Topics: Computer vision Machine Learning Data Science