In cooperation with twoday kapacity, Nykredit has developed a modern data environment in Azure that consolidates data and makes it easier to develop AI and Machine Learning models.
A new way of working with data
At Nykredit, they had a desire to use data more predictively. They were used to working with classic Data Warehousing, where they knew what to report on and what numbers to get out of their Data Warehouse.
Now they had to use data in a new way. And they realized they didn’t have the right platform to get to where they wanted with the project.
“When we work with predictive analytics and use AI and Machine Learning, it’s a slightly different way of working with data – typically more explorative and experimental in the initial phase of designing and developing solutions. That is why we needed some new processes. We wanted to make all of our data easily accessible to our Data Scientists – of course with respect for compliance requirements – so that they could test their business hypotheses and identify the data that provides both new insights and robust models,” says Per Vestergaard, Competence Manager of Business Intelligence and Analytics at Nykredit.
They chose to continue working with Microsoft’s Azure cloud data platform as they considered it the best fit for Nykredit. Rather than embarking on the project alone, Nykredit saw an interest in partnering with an external supplier to help them get started on the project. Here the choice as partner fell on Kapacity:
“We chose Kapacity because we had good experiences with them from a previous project. And then we knew about their vast knowledge of the Azure platform, where they have a good reputation,” Per Vestergaard says.
"We had an ambition to build an infrastructure on which we could scale and where we could meet the increasing demands we experienced. It was also important for us that we were able to gather our data on one unified platform, making it easier to use and reuse the increasing amount of data across the organization. It was a great driver for us to be able to create a future-proof data platform."
Jakob Hyttel-Sørensen, Supply Manager for the commercial part of BI & AA at Nykredit.
PoC provided the opportunity to test cloud capabilities in practice
Cooperating with Kapacity, Nykredit wanted to establish a data environment in the cloud that could support the development of predictive analysis models. They agreed to create a Proof of Concept with two specific use cases:
Establish a data platform in Microsoft Azure
Establish an environment for development and production, in which Nykredit’s Data Scientists can develop and produce AI models
The process began with some workshops between Kapacity and Nykredit, where Kapacity came up with different solution designs based on Nykredit’s needs. In this way, a common understanding was gained of how Nykredit wanted their future data and model platforms to be designed.
Based on a use-case driven goal, Kapacity and Nykredit’s employees then developed an initial version of the new infrastructure together. From the beginning, the process was organized with knowledge sharing in mind, so that Nykredit could subsequently begin the full production maturity, with all that includes of security approvals, role and process clarifications, etc. Training of competences internally at Nykredit was thus an important part of the process:
“Although we had used parts of Azure, our employees did not have a broad understanding of the services and technology that form the basis of a Data Warehouse mature for production in Azure. Therefore, we actively used pair programming with Kapacity’s consultants to ensure that our own employees gained concrete experiences with the platform while developing the various solutions. The process meant that we can now stand on our own two feet and continue the journey towards a setup fully mature for production,” Jakob Hyttel-Sørensen says.
twoday kapacity Cloud Proof of Concept (PoC)
A lot of companies experience an increasing amount of data and a business need, which can be difficult to meet with their existing platform. Here, a Proof of Concept can be a good starting point for testing the possibilities of the various Azure services. It helps the project get off to a good and fast start, you are able to clarify your needs and opportunities, and the prototype developed can most often be reused in the upcoming project.
Unified data is the key to predictive analysis
Nykredit mainly works with two tracks in predictive analysis:
Behavioral models, in which they work to predict customer needs and thus important sales signals. It can for example be churn models where Nykredit can predict which customers are at risk of leaving the business.
Decision models, where they work on automating manual processes. It can for example be automating document verification so that documents no longer have to be manually approved, but are automatically approved.
Nykredit has worked to create a data foundation that can be reused across models, but which can also be used more widely in the organisation.
“It has been important for us to manage our data because we see it as the key to working with predictive analysis and getting value from data. The key is the ability to quickly get from business idea to the first model results, because we do not have to start over each time,” Jakob Hyttel-Sørensen says.
Nykredit enters the cloud with Azure
Today, Nykredit has a hybrid Data Warehouse, something on-premise and something in the cloud. But according to Per Vestergaard, the goal is for the whole of Nykredit’s data platform to be in the cloud in the long run. They have decided to move forward with Microsoft Azure as platform, and they will use a large part of the other services that Azure contains.
There are several reasons why Nykredit has chosen Azure as their data platform, Per Vestergaard says:
“Because we are in the financial industry, there are a number of compliance rules and regulatory requirements that we have to live up to. And Azure is the best in the market to handle those requirements. In addition, Microsoft is large in Denmark, so there are strong competencies in the market that can help us develop Azure. Last but not least we tap into an extreme muscle of innovation, where thousands of people work to develop the platform so that it is constantly updated on the latest technologies in for example AI and Machine Learning. ”
Jakob Hyttel-Sørensen continues:
“We get a platform that addresses many of the things we have to deal with in everyday life, if we are to be successful. The goal is ultimately the business gains, but along the way everything from a scalable and reusable data foundation to operational considerations such as access control, version control and model monitoring are something we have to address. It’s all something Azure contains, and therefore we get a future-proof platform. ”
twoday kapacity has driven the project forward
Nykredit has been pleased with the cooperation with Kapacity and the knowledge that Kapacity’s consultants have contributed with. During the project, Kapacity has been good at sharing their knowledge with Nykredit’s employees, and pair programming has helped Nykredit gain an understanding of the structure of the platforms so that they have been able to take over and continue the work. In addition, Nykredit has been delighted with the energy that Kapacity has put into the project.
"We have clearly noticed that Kapacity’s consultants have been passionate about this project and have been very committed. It has given a lot of energy to the project, which has helped drive it forward."
Jakob Hyttel-Sørensen, supply manager for the commercial part of BI & AA at Nykredit.
Begin a cloud transformation small and centered around a use case
Today, Nykredit has further developed and produced their first models, and the outcome of the PoC became a decisive step towards a completely new platform:
“Working with cloud and cloud technologies has been a learning process for us in many ways. Of course, there are many ways in to such a process, but we have been really pleased that we have run it powered by a PoC with two specific use cases. This has made the dialogue more specific, which has been crucial for us, as we have been able to talk together across areas of competence. It covers the dialogue between Architects, Data Engineers and Data Scientists, but also in regard to stakeholders across Nykredit in general – including our lawyers and IT security department,” Jakob Hyttel-Sørensen says.
When asked what experiences they want to pass on from the project, they are not in doubt.
“For us it has been the best way to start small and center a project with new technology around one or more use cases. This way, you will gain some quicker experiences and clarify your needs. You also avoid losing your breath during the big journey that a transition into a cloud platform is,” Per Vestergaard says.
Nykredit is a company owned by an association of customers. Nykredit functions as mortgage lender, bank, insurer and estate agent. Their main activity is as a mortgage institution, where they account for 42% of the total mortgage lending in Denmark. They have roots back to 1851 when the mortgage began to take shape, but have been Nykredit since 1985, when Forenede Kreditforeninger and Jyllands Kreditforening merged.