What led you to the field of data science, and how have past experiences helped shape your expertise?
In school, I loved computer science, and I got involved in some fascinating data science projects around marketing and financial trends. This led me to applied mathematics and data science, and to my internship at Allianz Trade.
From there, I slowly began to take on more and more projects. Each one expanded my skills and understanding, and I’m leading projects and helping to identify questions that data science can answer.
What does your role entail and how does help drive innovation and collaboration at Allianz Trade?
My work falls into two categories:
The first is to be a bridge between the data science team and the business. I work closely with the credit assessment and credit underwriting teams to understand their needs. Then, I translate their challenges into data-driven solutions.
The second is to work on the solutions themselves: I lead the technical side of data science projects, planning the approach, developing models and making sure they are robust.
Data science is fundamentally about solving complex problems to help people. The tools we make and the problems we solve mean that our teams can work more efficiently, and that translates to better service for our clients.
What have you worked on recently that you’re particularly proud of? What impact did this have for our customers?
We have been recently updating one of our fraud detection models, a tool called Sherlock. Sherlock looks at all the context and past behavior linked to buyers (or our customers’ customers), helping us identify the ones that might not be legitimate. When it spots a request that corresponds with data from previous fraud cases, that is flagged. Tools like this help give our underwriters a complete picture before a decision is made, and these new updates incorporate our latest data to improve its performance.
We’ve also been working on a new project to enable more efficient decision-making on credit limit requests. Our underwriters get important information and context for their decisions from our statistical models, which use mathematics to create estimates of the relationships between variables. They differ from generative AI, which pulls information from a dataset to recognize patterns and create responses to prompts.
Statistical models can be helpful in making certain decisions, but understanding their outputs is a very skilled task. Our project involves looking at ways to make the outputs from the models easier to interpret, so that underwriters can access the information more freely and decisions can be made more efficiently.
Both these projects ultimately benefit Allianz Trade’s customers – from better fraud protection to quicker access to credit limit decisions.
What changes are you seeing in your profession and in the broader field of data science? What are the challenges or learning opportunities that come with them?
The data science landscape is evolving rapidly, particularly with the use of AI. And like with any new technology, it’s important that we identify its strengths and how to best use it in our business. But we also need to be clear on its limitations.
Data scientists today need to comfortable communicating insights and building trust with business stakeholders. In the credit risk domain specifically, there's growing demand for transparency – particularly with regulations and internal governance. And so, with the growing use of generative AI and large language models, we’re seeing new possibilities but also having to tackle questions around explainability, bias and responsible use. Humans need to remain at the center of our decision making, and our role is to help our colleagues use technology to do that with efficiency and confidence.
I’m excited to see how this space evolves, and I’m looking forward to exploring the possibilities as AI tools advance.
