Dmitry Kazakov, Data Analytics Team Lead at Kolesa Group, shares insights from the first Kazakh survey of data scientists.
In the photo: Dmitry Kazakov
Remember the popular phrase that Big Data is most reminiscent of teenage sex - everyone talks about it, but no one knows if it really exists. The same could be said about the market of data specialists (in Kazakhstan) - there is hype, and who is behind it (and whether there is at least someone there at all), it was not completely clear - neither HR nor managers , nor the data scientists themselves.
We spent , in which more than 300 specialists were interviewed about their salaries, functions, skills, tools and much more.
Spoiler: yes, they definitely exist, but everything is not so clear.
Nice insight. First, there are more data scientists than we expected. We were able to interview 300 people, including not only product, marketing and BI analysts, but also ML and DWH engineers, which was especially pleasing. The largest group included all those who call themselves data scientists - this is 36% of respondents. It is difficult to say whether this is covered by the market demand or not, because the market itself is only being formed.
The distribution of job levels is confusing - there are almost as many team leaders and managers as there are juniors. There may be several reasons for this. For example, a large number of small teams of 2-3 people, in which a middle or senior specialist can be the leader.
Another reason may be the chaos reigning on the market so far in terms of standards in the distribution of roles and functionality. Team leaders are sometimes appointed by those who simply work a year or two longer than others, without reference to the level of skills and knowledge. We see this in the distribution of functions by position - 38% of managers and team leaders are engaged in pre-processing and another 33% in basic statistical analysis.
Here we asked respondents to subjectively assess the level of analytics in their companies. If you look closely, you can see that 10% of respondents who work in analytics departments of 2-3 people consider themselves to be “advanced”.
What is "advanced level"? The BI system works great. There is DWH and Big Data. A/B tests are conducted regularly. There are working ML and DS systems in production. Decisions are made based on data. The department of work with data and Data Science is one of the key ones in the company.
All of the above is almost impossible to achieve by a department of 2-3 people. I think that such a result of the survey is a small growing pain - the guys still have no one to compare themselves with in order to determine their level more objectively.
As expected, data scientists spend most of their time not on super complex mathematics or engineering, but on pre-processing, uploading, cleaning data. In each specialization, we see preprocessing in the top 3. But we rarely see complex things like developing ML models or working with Big Data in the top 3 - only among ML and DWH engineers.
There are also a couple of sad insights. 40% of the tasks are set by specialists themselves. So far, only top unicorn companies in Kazakhstan have tasted the benefits of working with big data and learned how to do it competently. They broadcast to the market that Big Data and Machine Learning are cool, and the second tier is following, but they don’t always understand how working with data works. Therefore, we see that specialists set tasks for themselves, and business does not always know what it wants.
I was surprised that 20% of specialists do not even know if their company has a Data Warehouse. Yes, and database management systems are not so good - 41% use MySQL, and another 34% use PostgreSQL. What can it say? They work rather with small data.
In the question about storage systems, we again see MySQL and even (!) Excel. But this may indicate, for example, that most companies simply do not yet have a request to work with big data.
Here again everything is ambiguous. In general, salaries were slightly lower than I expected.
Personally, it's hard for me to imagine an ML engineer who is ready to work for 200 thousand tenge - probably a trainee. Either the competencies of such specialists are very weak, or it is still difficult for companies to adequately evaluate the work of Data Science. But perhaps this also suggests that the market is still at the very beginning of its maturation. And over time, the level of salaries will be established at a more adequate level.
Source: habr.com