Can someone suggest projects in big data

Chief Data Officer at the age of 29 - the interview with Nikita Matveev

In the data driver interview with Nikita Matveev, Chief Data Officer of S7 Airlines, you will find out how you can build a team of data scientists in a short time, what makes a successful team, how you design data products and what experience Nikita already has with our data driver method , the data strategy design.

Thank you Nikita for letting us do this interview with you.What is yourRole as Chief Data Officer and what does S7 Airlines do?

S7 Airlines is the second largest Russian airline and covers around 20% of passenger air traffic in Russia. Ultimately, my job as Chief Data Officer is to get the maximum value from our data and thereby boost business.

How did you get into your role as Chief Data Officer (CDO) at the age of 29?

Talent, talent, talent,…. (laughs) In fact, I think I had the right background for this position. For one, I have previously worked as a strategic consultant and therefore management experience. On the other hand, I have a degree in physics. This combination helps enormously in the role of CDO. In 2016, at the beginning of my career at S7 Airlines, there were many big data projects in connection with machine learning on the program. This area developed extremely quickly and we decided to set up our own data department.

How many people are working on your team now and how is it put together? What are your success factors?

Overall work 65 people on data projects as business analysts, data scientists, database administrators, developers and product owners. We also have less specialized ones Data Analysts, Data Quality Analysts and Data Developers included in the data office. By combining different competencies in one department and our advanced data maturity, we can easily implement solutions.

Our success factors:

  1. As already discussed in the seminar today [The interview was conducted on the sidelines of the Data Business seminar.], We focus on “Holy Grail” projects. In other words, use cases that seem difficult or even impossible to master, but sound very good and on which a lot of attention is paid company-wide. These projects should be pursued in order to motivate employees and to get support from outside.
  2. Then you should find out which work packages can be implemented and then split them up accordingly.
  3. And finally, risk management and business analysis are very important in order to be able to evaluate the feasibility of projects in good time.

In November you gave a presentation at the Predictive Analytics World for Business conference in Berlin on the subject of “Brick and Mortar of a Data Science Product”. What do you think the bricks are and what is the mortar? And how do you define a data product?

In my opinion it is a data product an IT system that is integrated into an ongoing process and provides added value.

I would say there are three bricks and one mortar. The first brick is the product definition: There are many ideas for data products. In order to make the right choice, the problem to be solved, the stakeholders and the customers should be analyzed.

The second brick is the team and the process. In order to set up an IT system, various skills are required from data scientists, engineers, DevOps specialists and business analysts who work closely with the specialist department. This also includes certain processes for control. We use Kanban, for example.

The third brick is the business analysis. As already said, a data product is built into the company process as an IT system. Hence, the best approach is to identify the right requirements and risks.

The Mortar is the data ecosystem. If the foundation is not in place, it can take years to develop a data product. Projects such as a data lake or a data catalog have so far been used more than terms in marketing, but if they are set up correctly, they help with data product development and can reduce costs.

We first met in May when you took part in our data driver seminar “Data Thinking”. Today you attended our data business seminar. Why did you come here for the first time and what did you expect?

I was in the process of building a data strategy and thought to myself that someone would have done that before me. While googling, I discovered the data driver website, with interesting material and the corresponding canvas. That was exactly what I was looking for.

And have you already used our canvas tools at S7 Airlines to develop a data strategy? And what was your experience with it?

Yes, I have tried the method with different departments and you have been very useful. We even conducted a full day workshop with one department and developed an analytics strategy. They were very happy with the approach and the results.

What is the main benefit of using the canvas tool and the data strategy design approach?

I would say there are three reasons for using data strategy design in projects. It helps:

  1. identify more risks and save time.
  2. to establish better communication in the team.
  3. build a better structure. And the materials created can be used again and again in the future as a basis for other applications.

What are your tips for someone just starting out with data strategy design?

First I would check out the data driver website to learn more about the method and canvas.

As a second option, I would suggest taking part in the data thinking seminar to find out more about the process, the organization, communication methods and the appropriate tools.

After participating, I would suggest planning a PoC (Proof of Concept) workshop in your own company. Since a full-day workshop can be too strenuous to start with, I would recommend going through parts of the design kit with various stakeholders. For example, I've done this with different departments. We also transferred all of our projects to the analytics maturity canvas, which was good training and we were able to create a clear roadmap.

Thank you, Nikita. May the data be with you!

Alternatively, you can watch the interview as a video on YouTube: