We were at the Data & AI Warsaw Tech Summit 2025

Some of you may have heard about our debut as speakers at last year’s Data Science Summit. If not, check out the article on our Tech Blog, where you’ll find a summary of our talk – A remarkable journey – from DataLake on AWS to Databricks.

As you might expect, it was a very interesting experience for us, which is why we wanted to present again at this year's conference. We chose the Data & AI Warsaw Tech Summit 2025. On stage, we were: Jacek Cieślak - APEX Service Manager, Robert Marek - APEX Engineer and Mateusz Wujec - Data Engineer, sharing our experiences in optimizing fraud detection in financial institutions using AI and organizational tools. Our presentation focused on the use of advanced technologies to detect suspicious transactions, as well as the optimization process of real-time transaction monitoring.

Our presentation began with a discussion of the importance of marking suspicious transactions and presenting the basic business logic applied in banking systems. We covered various examples of transactions that are monitored, including those related to cryptocurrencies, high political risk, or money laundering through offshore institutions. We emphasized how critical it is to detect such transactions in order to prevent terrorism financing and other financial crimes

.To address the challenges associated with analyzing millions of transactions, Raiffeisen Tech uses Apache Pulsar for real-time data streaming and Apache Flink for efficient data processing. Transactions are analyzed in real-time, and the results are stored in databases like Databricks, enabling subsequent verification and analysis by the compliance team.

Another challenge discussed was the issue of knowledge sharing within the organization. Due to the complexity of systems and specific business knowledge, it is often difficult to gain a full understanding of the required context. In response to these difficulties, the speakers presented an AI-based solution that supports employees in accessing knowledge. Using AI, we created a chatbot system integrated with the Confluence platform that facilitates easier information search and processing in the context of fraud detection.

The chatbot works by searching the knowledge base, utilizing a vector model to accurately match results and provide the most relevant information in response to user queries. This system is integrated with Microsoft Teams, enabling users to quickly access the necessary data in an intuitive, conversation-like manner, as if speaking with another employee.

To conclude, we emphasized how such technological solutions, combined with advanced analytical tools and AI, help optimize the fraud detection process and support effective compliance, while also providing greater operational efficiency in banking.

Our appearance at the Data & AI Warsaw Tech Summit 2025 was not just a technical presentation — it was also a great opportunity to showcase how artificial intelligence and modern tools can support regulatory compliance and enhance financial security.

If you're curious to learn more, check out the recording of our session. We also encourage you to explore the English version of the presentation we discussed during the event. It’s worth a look — you'll find plenty of valuable insights and inspiration on how AI can be used to detect fraud!