Crunching Data to Fight Financial Crime–The Expert’s Take

My Data Guest — An Interview with Emil Cimpean

Rosaria Silipo
Low Code for Data Science
6 min readApr 17, 2024


My Data Guest — An Interview with Emil Cimpean.

It was my pleasure to recently interview Emil Cimpean as part of the My Data Guest interview series. Emil works as the head of the Financial Crime Compliance at ING Romania. He has been working in this field, fighting financial crimes, detecting frauds and money laundering activities for many years now.

With Emil we talked about how data science can be used to fight financial crimes, we saw what are the features of KNIME that come in handy to an AML (anti-money laundering) expert, and finally he shared insights into the future of financial crime prevention with AI.

Rosaria: Let’s start with your job, Emil. I’m sure everyone’s curious. What exactly do you do? I picture you out there chasing down the bad guys. Are you? Give us more insight into your role.

Emil: I’m the Head of the Financial Crime Compliance Customer Activity Monitoring at ING Romania. Contrary to the Hollywood image, we’re not out there chasing bad guys in the streets. In reality, most of them are hidden behind screens, using laptops or smartphones. Our role is to be part of the mechanism that combats criminal activities.

We use data science to monitor and detect potential money laundering cases that are then reported to the local financial intelligence unit. I’m proud that my team and I provide a safer environment for our customers and, in general, to society as a whole.

Rosaria: Can you elaborate on how data science fits into your job?

Emil: Data science is crucial in our day-to-day operations due to the constantly evolving technology landscape and the increasing volume of data we need to analyze. We simply can’t afford not to leverage data science if we want to stay ahead. Understanding patterns and anomalies in transactions, which are key in detecting money laundering, relies heavily on data.

Rosaria: Could you provide some typical use cases where data science is utilized to detect financial crimes?

Emil: Data science plays a critical role in identifying transactional patterns and detecting abnormal activity in transactions. To conduct high-quality analysis or investigations, we need to integrate and link relevant data to paint a complete picture. Visualization tools also come in handy, helping us understand cases better and identify all participants involved in complex schemes from just one initial transaction.

Rosaria: How long have you been using KNIME?

Emil: I’ve been using KNIME for about seven years now. I was introduced to it by a colleague during a workshop focused on fighting cross-border VAT frauds. This colleague had used KNIME to present his approach to a fairly complex data analysis. I was positively surprised so I decided to incorporate this tool into my work and projects.

Rosaria: As a seasoned KNIME user, how does it aid you and your team in your work? What does the classic workflow look like?

Emil: I use KNIME on top of other tools to improve the overall analysis and investigation processes. Sometimes the tasks you need to perform are better suited for a software like KNIME, compared to others like Excel. We still use Excel as a starting point but there are visible limits to it so KNIME becomes our go-to solution for handling complex data scenarios. When a project gets more complicated, KNIME’s seamless project sharing facilitates team collaboration.

An interesting example of how we use KNIME is to detect money mules, i.e., individuals who are paid a small portion of illegally acquired money to transfer it on behalf of others. In my team, we leverage KNIME’s network analysis and clustering features to efficiently pinpoint participants in such illegal schemes.

Rosaria: What are the features of KNIME that you find most useful in your work?

Emil: I particularly appreciate KNIME’s text mining capabilities, network visualization options, and its speed in processing large volumes of data. Its automation features are invaluable, serving as a reliable assistant in recalling data processing steps and facilitating documentation for future reference.

Moreover, its visual nature simplifies the adaptation of existing workflows (or segments of them) to new tasks. When copy-pasting segments of a workflow, users clearly see what they are transferring, enabling a better understanding of the process at hand.

Rosaria: What about your top three KNIME nodes?

Emil: If I have to choose three, I would say:

  • The Joiner node, that just makes the VLOOKUP function from Excel look so hard.
  • The GroupBy node that I always use, especially its “unique count” option and the “list” aggregation option.
  • The Duplicate Row Filter node is also very useful in my case, because I often need to check for duplicates or unique values.

Rosaria: In your opinion, which part of the data science lifecycle is the most challenging to implement and maintain?

Emil: You might have noticed that my top three nodes all revolve around the initial stages of a data project, when you collect and preprocess your data. This isn’t by chance; I firmly believe that this phase is the most challenging one, given the diverse sources and data quality involved. Handling issues like data accuracy, missing values, and integration can be quite complex, but tools like KNIME streamline these processes compared to more traditional methods.

Rosaria: What are your thoughts on AI? Do you see it as promising?

Emil: AI holds great promises, especially in streamlining routine tasks and improving efficiency. In the field of AML, AI can be leveraged to recognize patterns and outliers in transactional behavior, so it’s a matter of time to see its implementation.

However, it’s crucial to recognize its potential misuse by criminals in illegal activities. The current advancements in AI and image processing can be exploited by criminals, for example, to produce high quality forged documents, a common occurrence in financial crimes scenarios.

Regulation and responsible use are essential to benefit from AI effectively.

Rosaria: Have you encountered difficulties in hiring data scientists and KNIME experts?

Emil: Yes, finding skilled data scientists is already challenging, but finding KNIME experts, even more so. While AI may address some tasks, expertise in data analysis remains highly sought after. That’s why I always encourage my team to attend KNIME courses and actively engage with the KNIME community.

Rosaria: What tools do you believe data professionals need to know to work effectively in your group? Python? KNIME?

Emil: In my group, familiarity with KNIME is essential and it’s even more important than proficiency in Python or other programming languages. While expertise in Python or similar environments remains valuable, particularly for projects of greater complexity, KNIME stands out as our default tool due to its intuitive and user-friendly interface.

We also use SQL to generate reports, complementing our data analysis workflow effectively.

Rosaria: We’re reaching the end of our interview. Before we say goodbye, we have a couple of more questions.

Sometimes mistakes are good because then we can learn more. Tell us about the biggest mistake you have learned from.

Emil: It certainly has to do with data and the most valuable lesson learned is to never start working on a project until you fully understand the data and verify its quality and completeness. The risk is that you would end up spending twice as much time identifying (and fixing) the data issue.

Rosaria: Now the opposite question. Tell us about the most successful project you have ever worked on.

Emil: This is also related to data and data science. It was a project with more than 50 entities involved in a VAT fraud scheme where I succeeded in tracking down the products moved through many companies and to detect their real path, which was different from the money path that went through the banks. Such discrepancies were made on purpose to mislead the investigators and make the detection of the real movements of the goods harder.

Using formulas, keeping track of product quantities, and creating automated checks and validations allowed me to successfully identify the fraud and report it to the competent authorities.

Rosaria: Thank you, Emil, for sharing your insights with us today. How can our audience reach out to you?

Emil: I’m always open to professional discussions, and LinkedIn is the best way to connect with me.

Watch the original interview with Emil Cimpean on YouTube:



Rosaria Silipo
Low Code for Data Science

Rosaria has been mining data since her master degree, through her doctorate and job positions after that . She is now a data scientist and KNIME evangelist.