Eben Serfontein, Credit Risk Manager with U.S. Bank Europe, was a recent joint winner at the IOB Future of Finance Awards 2025 for Risk Management.
Eben was recognised for his leadership of the design and development of an innovative risk measurement model that has transformed how U.S. bank measures liabilities across its payment’s portfolio. His data-driven approach has helped protect customers and strengthen improve the organisation’s risk management decision making.
With a background in engineering and extensive expertise in big data analytics, Eben is passionate about the opportunities that new technologies offer to financial services organisations, customers and our society.
We speak with Eben about data-led risk management, the importance of stakeholder engagement and the power and potential of AI for fraud protection.
I’m inherently driven by wrestling with unsolved problems and working through the unknown to achieve solutions. This led me to studying Electrical and Electronic Engineering. After working a few years as an embedded software engineer, I had an opportunity to apply a similar skillset to quantitative modelling at a large bank in South Africa. The mixture of data wrangling, mathematics and finance was an intriguing recipe and brought an endless number of opportunities such as improving strategies for capital management, portfolio credit risk management, pricing initiatives, and financial forecasting.
Transitioning to Ireland, I was able to pivot the same skillset into new types of portfolios that has kept the journey of continuous learning alive. I enjoy the balance being sought in risk management of solving solutions that are both in the interest of customers and provide growth within the parameters of the risk appetite of the bank.
There is a certain amount of nuance to how credit risk crystallises in a payments portfolio, especially within acquiring business models. Regulations centred on credit risk tend to be shaped by traditional credit products, and established measurement techniques therefore aren’t as established for other types of credit exposure. The project was an opportunity to take the core guidance from regulatory and accounting frameworks and then model the way that off-balance sheet risk is assessed quantitatively in alignment with those principles, that also delivers a logical result that makes sense to a subject matter expert.
There were challenges in research due to limited directly applicable text, as well as building new data structures and statistical analysis. The unique nature of the project also necessitated significant stakeholder management, as I wanted to ensure all the stakeholders needs were met and that progress was being tracked effectively. When treading new ground, it is essential to be transparent but concise in communicating with stakeholders both in terms of the key message we were communicating and what we needed. This allowed leadership to make informed and collective decisions on the direction of travel and ensured that the solution stayed on course.
Data on its own is just ones and zeroes. But harnessing data can extract critical information when the right questions are asked, which can then be churned and converted to knowledge through quantitative techniques. It is important that the data is managed properly to remove noise that could mask the true picture or even result in false conclusions when not mixed with business sense. But data that is respected and well managed, provides a concrete point of inference to teach us about behaviours that are sometimes less obvious or even intuitive from a ground level perspective. This allows us to detect changes in a portfolio that can threaten the risk appetite goals of the bank at an early stage so that the risk can be monitored, communicated and managed gradually.
Sometimes the pace of change is more drastic, but even then, having a means of identifying the critical components of stress in the portfolio is all helped by effective data infrastructure. Even stepping beyond inference, by the knowledge gained, effective predictive techniques are possible only through strong data sources, and predictive techniques enables us to plan and strategise based on the most likely outcomes, while also keeping an eye on the unlikely ones. Without data, such estimates can more easily be influenced by human emotion and desires that can mask the optimal balance in risk.
I have investigated how generative AI can be used as a tool to depersonalise data that is used in fraud detection models. The concept is that the deep learning algorithms can replicate the training data in such a way that is still contains the core behaviours that a fraud model would require to accurately predict which transactions are fraudulent. With the obvious applications of AI in the fraud prediction side, this research showed that generative AI can also be used to protect privacy of consumers by depersonalising the stored data that fraud models are trained on, thereby limiting the risk of leaked sensitive information. This echoes the principles of GDPR related to data security by design.
In a technology dependent environment, taking learnings from the wider technology industry could be helpful. Many technology companies would cultivate innovation through dedicated days for hackathons and other initiatives where employees can explore innovative ideas without being directly attached to their daily work. This develops a mindset of creative thought that sits in the back of the mind even when one returns to the core job tasks and initiatives. It also provides a melting pot for new concrete ideas to come to the fore that could keep the company a step ahead.
In the execution of your role, you can become very focused and lose perspective on the impact of your efforts. At the same time, there are many people doing amazing work that combines to make the company great, but it is very motivating to know that your involved in delivering results that matter to the organisation.
As technology evolves and accelerates the wave of change, it is easy to jump onboard and lose sight of critical safety nets that ensure clients and even the wider public are protected from undue harm. Using technology and data for good should be central to the decision-making process, to ensure what seems like a forward leap isn’t accommodated by unintended consequences.
Eben Serfontein was joint winner of the IOB Future of Finance Award for Risk Management. Learn more about the IOB Future of Finance Awards 2025 winners here.