CIO of Raiffeisenbank Andrey Popov - about IT strategy, focus on Agile, DevOps and the open source software

Andrey Popov: There is a beautiful answer that the advanced players in the market have a vision of IT as business, but not as the additional, servicing division. It sounds well, but, I think that so far not everywhere and not always. Eventually, it not about bank, and about any service organization. Its technological effectiveness is decisive factor success and development now. Banks are those at whom IT and business of each most well other understand, are most integrated.

If to look for the last 10-15 years, it is well visible how the role of IT turns from an authoritative source of information on how to do or to do specific technical solutions, in the one who helps to create solutions, products and services for end consumers, and not just to perform some set of operations according to the list. Absolutely precisely I can tell it about Raiffeisenbank. At us the role of colleagues from IT who help business to understand, where to us to develop in terms of a technology platform is very high. Also the role of the divisions connected with internal R'n'D, innovations is high at us: they help with search of new solutions in the market too to interweave them into future products of bank. It was not earlier.

Andrey Popov: Our chairman of the board is very close to questions of what he wants to construct the organization in terms of both digitalization, and the relation of people, and work of people in bank. A question not so much in the digitalization, how many in how to construct such organization what should be people who in it work and as to achieve such status. It, I think, point number one in its agenda. Therefore support of everything that is connected with digital transformation in bank, and even occasionally the call on a subject as to us to make better and more, from Sergey is broadcast constantly. And, the most important that boards are up a unison to standard that it is really important, is necessary. To me in this plan to work very comfortably.

Andrey Popov: I would not begin to select something specific. All say about Big Data, use of unstructured information, machine learning, artificial intelligence much and about how all this helps clients and employees of the bank. All this is beautiful, but not the fact that it is to straight lines decisive factor that, this or that bank is doomed, or not doomed.

It seems to me, a question not in some specific technology, and in approach to how we work with the client and why we do it. It is important that the end product was convenient for the consumer. Convenience, availability, an opportunity to use this or that solution or service, anywhere and as the client – here wants that is really necessary. It concerns clients and physical, and legal entities. Therefore technologies which move us to it, in a prize.

For example, it is possible to refer everything that is connected with machine learning because it helps us to formulate that really valuably to the client or to try better to understand it to such technologies. But in the same way, for example, design thinking allows to improve communication with the client though it is not technology, and it is rather, model of building of relationship.

Andrey Popov: It is, for example, mobile development. Now all of us carry with ourselves gadgets, and we do not want to be tied to the stationary PC. Here it is possible to carry everything, as for experiments with Big Data, with the analysis of a customer behavior on social networks, attempts to get access to sensors of Internet of Things, geolocations. It is possible to look also at something other, at wearable things: for example, hours as a source of the additional information for targeting and tagging of people. Also the biometrics has perspectives. It - what we invest in.

Andrey Popov: Quite recently we just spoke on this subject in the command and with a number of board members. Somewhat, existence of separate IT strategy in the organization is an anachronism, considering as the role of IT and business in the organization changes. Nevertheless, we have a strategy. Even if to call it not the strategy, and the route chart of development of technological capabilities of bank, it should be present at some type because helps to consolidate information on what needs to be made for business in the field of IT.

The current horizon of strategy at us till 2020, now we pass through a cycle of its updating. Last year strategy was accepted, year passed, now it is necessary to introduce certain amendments.

At the same time how IT strategy is formulated in Raiffeisenbank, does not look as the typical IT strategy of banks if to compare to what I saw. It does not affect the level of specific projects, specific objectives, and speaks more likely about the directions which to us are interesting and important, and not only in terms of technologies. In addition to technologies in which the bank wants to invest in IT strategy we speak also about culture much from that point of view that the IT organization was a part of business, about the culture of business, about culture and philosophy of Agile. Also we speak in it much about architecture how we change an applied landscape to reach business objectives. Also subjects of budget, financial performance, development of people are touched.

The technology component of strategy is directed to promoting Agile distribution as philosophy, culture and methodologies of development and implementations, and solutions, development of products which corresponds to this culture. It both Scrum, and Kanban for these or those divisions.

Everything that helps teams to work thus, is priority. Here also the microservice architecture, focus on solutions with an open program code enters. And to give the chance to commands to make these or those changes which allow them to be more effective there is serious focus on practicians of CI (Continuous Integration is continuous integration) and CD (Continuous Delivery and Deployment is continuous delivery and deployment), DevOps to help cross-functional teams to work effectively and so on.

Andrey Popov: At the moment open source at us is available to use practically at all levels of systems, except mission-critical (crucial), just owing to the fact that there a number of vendor solutions which do not support open technologies interesting to us is used. For the rest, in principle, on any of levels of criticality of the Open source software maybe in certain cases even it has to be used. The greatest, probably, success which we had – everything that is connected with Apache: servers and application servers, TomCat and another.

Also we very successfully have use of PostgreSQL as the alternative database which in certain cases perfectly replaces proprietary solutions which were till this moment, for example, Oracle and DB2. But, by the way, the SQL server of Microsoft normally solves these or those problems. Especially, if it is the homogeneous environment in Windows development.

The open source software is used also at the level of business solutions, frameworks. As WorkFlow and BPM we actively began to use Camunda. Everything that is connected with a stack of Big Data - it is Hadoop.

As the operating system we widely use Linux.

Andrey Popov: We began with internal things – bases of the Service Desk system. Now we look at that to transfer a part of business applications and the front systems to PostgreSQL, to use it as base of storage for VRM of solutions. Where we optimize use, let us assume, of DB2 or Oracle as platform base, there PostgreSQL very well rises just.

Andrey Popov: We have no such rate. In the current phase of development when we at the level of business set before ourselves ambitious goals when it is necessary to carry out serious transformation of a business model, to shift it more in "digit", it is better to understand the client and to adapt under his requirements, it is impossible to speak about decrease in operating costs of IT. It would be how to cut chicken who lays gold eggs.

Certainly, it does not mean that there is carte blanche – spend how many you want. Raiffeisenbank always pragmatically treated how and what expenditure become. Therefore there is a close attention for my part to how IT budgets are constructed, and from financial directorate and, in principle, from board. But a talk that to us it is necessary to reduce, for example, by 10% IT costs in 2018 no. And if to look at operating costs, OPEX of 2018, then they, most likely, will even be higher, than OPEX of the 2017th. At the same time there is a structural change: we reduce capital costs just owing to the fact that at the end of last year and at the beginning of it there was a serious process on insourcing of a number of areas.

IT of CAPEX this year we lower by 2%, at the same time the general budget of IT will rise more than 25% without depreciation.

Andrey Popov: For example, we transferred to bank of staff of Development center and maintenance of information technologies in Omsk. Earlier we started this center together with the partner, but at the end of last year transferred to bank. It is technical support of key services of bank, product development, management of large-scale technology projects of bank.

Andrey Popov: I think that it is not necessary to oppose a fashion and advantage. On the one hand, apparently, that it is a fashion. All are engaged in a blockchain, try to find some real case which will issue something phenomenal. But, on the other hand, it is not a mainstream, and a stunning case where the blockchain would solve all problems which could not solve before, not so far. And I am not sure that it will be in the near future.

I think that the blockchain is story about how methods appear to transform the settled business models for other society in which there are many things which algorithmic are solved using systems, agents, bots, etc. And the blockchain here just becomes a certain blood rule if to draw a parallel with an organism. The blockchain is actually the fact that allows to construct interaction between the different organizations, agents in some relationship. As the technology, as the platform it allows to build much more automatic and algorithmic relationship: using smart contracts, different oracles who look in these or those gateways and react to these or those changes of events. From this point of view there is a high potential.

Of course, there is also a huge number of problems, for example, with a performance, at information transmission rate which, most likely, will be solved.

Andrey Popov: From practice we liked experience of last year together with MegaFon and NSD. I think, we will develop this story further as the tool was rather simple. And with colleagues from NSD it is interesting to us to try more difficult tool, perhaps, and with coupon dates, and with difficult lifecycle, and not with only one buyer as it was in a case with the previous case. Trade and financing – the beautiful candidate to try a blockchain. I think, we will have some stories here in 2018.

Corporate clients have huge interest in solutions on the platform of a blockchain. Often, however, their desires is much better, maybe, are implemented not on a blockchain, and on systems with open API that we also explain to them. And we with it are ready to help clients too.

Andrey Popov: Machine learning – it is very interesting. When we began to speak about machine learning in bank, it was talked of using insights of our client information at which we did not look earlier, and to attract startups or such coryphaeuses like "Yandex" who are engaged in it. We have rather big division - risk managers who, in fact, always use these technologies, build models, and they offered us the abilities. It became promoting of existence of capabilities and abilities in the organization of which nobody even thought earlier. Colleagues actively participated, competed who will better construct some model.

From this we had a number of such interesting predictive stories for CRM and for assessment of internal operational efficiency. And I am inexpressibly glad to it because I opened huge internal potential in the organization.

In terms of specific cases, we started the project with ATMs on demand forecasting on cash. We worked hard around the analysis of texts which at us both in service-desk, and in preparation of predictive base for agents that they could work requests of clients quicker. So far it at the level of tests. In this case I see big potential advantage - to help the specialist with information center, proceeding from that description which the client gives, quickly to pull out the most probable further ways of development of its request.

There is an idea also to construct using machine learning as much as possible systems for self-service of users inside. Many the arriving requests can be solved without person. We work to construct a system which would be as the certain conductor for users helping to solve this or that problem. Perhaps, in the form of the chat-bot.

Andrey Popov: The ATM has, in fact, two statuses, two very important parameters. While the client approaches it, his his basic function – to give money, and in it there has to be enough money in order that to service the client. On the other hand, from the point of view of bank, the cash loaded, and the ATM is a dead load. It is the money excluded from turnover. On the other hand, around them there is transport cost connected with collection. If you put too little, then you to the ATM often go to load it. If you put too much, then you can have situations when too much money leaves turnover.

Also there is a task - correctly to build up, balance model for each specific ATM. At the same time it is necessary to take into account that in addition to our ATMs there are still partners and to estimate influence of the partner ATMs standing nearby on our ATMs. For this purpose the model using machine learning where is defined how many money into what ATM to load how often where to go how to construct an optimal route is under construction. This combination of transport and packaging tasks and optimization of cost of cash which is in ATMs.

Andrey Popov: Yes. We only because of it also did the project to liquidate losses. Now we save about 10-15% of that how many we spent for it earlier.

Andrey Popov: Thanks to our practice in this area we have a very good reputation in the market of a fintekh what I am awfully glad to. There are several stories. First, it is important to look at what occurs in the market because itself it is not always possible to ingenerate interesting new things. On the other hand, I is initial when we started our innovation program, looked at option more likely, conditionally, an accelerator, than some venture fund. It is interesting to us to see what is in the market and to do with developers pilot implementation, to estimate as far as this or that hypothesis works. It also urges on colleagues in business to leave from a paradigm "let's make as at company X, maybe, it will be pleasant to our clients too".

Work with startups, with fintekha in this plan – a great way to make 50-60 experiments in one year instead of 5-6, having saved money, human resources, time. In turn, startups are interested in such cooperation, work with large bank and resources which we offer is interesting to them. And it is interesting to me to approve them. Fine symbiosis turns out.

We began in August, 2017, and since that moment of 16 cases worked completely from beginning to end. It was preceded by receipt of hundreds of the different ideas to us from which from the final we selected 50-60. From remained as a result of 16 we 4 we take in hand – we try to implement them in commercial real operation in bank. And now at us, 5 or 6 hypotheses in the current study.

When we began works with startups, in bank began to ask me questions why external cooperation, "why not with us is selected from within?". The answer simple - with external it was simpler to us to start. In 2017 we started also an internal innovation process in which we select two directions. One is connected with basic researches - with a blockchain, speech analytics, neural networks, etc. Perhaps, a story with Internet of Things, biometrics and so on will go there. This group conducts work together with Fintekh association of the Bank of Russia.

There is second, more applied group. It is engaged in researches which are turned on improvement of internal processes and approaches. It not about something absolutely new and very perspective, and trying the clear existing tools on the processes existing in bank.

Periodically we discuss also history around the venture funding. In terms of the current model of functioning of business, we do not consider yet creation of venture fund which would invest in these or those startups. But our colleagues at head office stare at this perspective more. Perhaps, we will work in partnership with them if we see any interesting startups in the Russian market.

Andrey Popov: The main effect that culture and philosophy changes. It in what each other people understand better better see what is important that is necessary. They can make more decisions independently, demand less involvement of internals of the power, institutes of declaration of will.

Last month we had a group meeting in Vienna, and we three together - Sergey, the chairman of the board, my colleague Geert Hebenstreit who is responsible for financial directorate, and I - told about where we are now. And Geert gave very interesting example. He said that it has a command in the operational block which sent the requirements to an external command earlier, from requirements queue was built. And often there were disagreements who not so made something with analysis of the reasons. The IT had no complete joint with the client. And from the middle of last year they began to work as the cross-functional team distributed in three places. There is a customer of this story who sits in Yaroslavl, a part of a command sits in Moscow, and the most part of analytics and development – in Omsk. They cross functional and cross territorial.

Because colleagues are involved in very close communication now – that is necessary why and why, the IT command saw what effect is rendered by their work on processes in the operational block. In turn, children the operational block began to understand why these or those things can seem simple in terms of "I want", but at the same time difficult in implementation. As a result they have less conflicts and the climate in this cross-functional team is much better. They understand why. And this the most important that it is possible to receive from Agile as from philosophy, from culture. It is impossible to implement it because culture is not implemented. Culture sprouts.

There was a time when we hoped that we will just provide trainings, will train all people, they realize as far as it is healthy, and will do all this in a new way, at them everything will turn out, they everything will be glad. But as a result we understood, it does not work. Therefore the last 1.5-2 years we systematically implement techniques connected with the organization of development process. The main of them is Scrum, just owing to the fact that it is simpler to explain, inform of it both IT commands, and business divisions. But there is a number of commands for which Scrum as it is very tied to regularity of an exit, to sprints in 2-3 weeks, can be improper. Especially for those who are focused on the flow which was less regulated on terms. There something else is used – for example, by Kanban.

At the different level of a maturity in terms of processes, both IT, and business, about 30-35 commands from total number of the commands created around different areas, business processes or systems work at 70-80 for us on Scrum now. In some there is already full-fledged cross-functionality. They are kollotsirovana and sit together with colleagues from business. Somewhere it is slightly more history IT, and business still slowly gets. And I very well see how the relation at our business partners changes as they begin to understand why it is necessary to explain why it becomes as far as it is important. I see, how more seriously colleagues from IT begin to treat what they do.

Andrey Popov: In all directions of business we have Scrum-commands: and in corporate business, and in retail. In retail began the first, and passed everything that is connected with card processing, colleagues from in crediting, pipelines some of such pioneers too to Scrum 1.5-2 years ago. In corporate business everything that is connected with development of acquiring, Internet bank for legal entities, already on Scrum.

Andrey Popov: My main KPI, and we broadcast it on commands which work in Scrum-methodology, in functional product teams, is focused on two things – customer satisfaction and a financial result. All this what their KPI comes down to. It seems to me, it much the best criterion, than quantity of the story points displaid in commercial operation for unit of time.

Of course, I expect that the speed of implementation will be more. But I consider that just it is not a final criterion of success. The satisfied client, growth of NPS (customer loyalty) on these or those products, channels and a financial result which the bank or a product command receives is more explicit external criterion. Inside is a criterion of the involvement and satisfaction of those people which in these commands work.

At the same time in the organization which is constructed around cross-functionality, is focused on the client, a story about projects in principle disappears because you do not do projects for improvement of these or those products. You is permanent, day by day you improve what you provide for clients. The project is something that has the beginning and the end. Very much it would not be desirable that relationship with the client had the end. The product should change constantly. New elements of functionality should appear constantly as the answer to this or that need of the client.

Andrey Popov: In catastrophic (laughs). I consider that success at a cross-functional product Agile-command or a Scrum-command without strong the practician of DevOps is impossible. DevOps, in fact, is that base at which such commands can work. There is no DevOps, there will be no Agile. DevOps is the Agile introduced more on the bekovy party. Practically in all commands in a varying degree practicians of DevOps are implemented.

Target status to which I want to come are completely cross-functional DevOps-commands in those divisions where it makes sense. For certain there will be some small set of stories where it is less comprehended, than in others. But I want to implement the principle "you build it, you run it" (that constructed, operate), it very much is pleasant to me because it means responsibility for what you created. Then there are no stories that, some something was nakodit, and another now on the shoulders to carry it.