From astrophysics to AI-supported skin research

My fascination with science began early in school. I was always curious to understand what lies beyond what we can see at first glance – to uncover the mechanisms and beauty hidden beneath the surface. I was particularly captivated by Jupiter, with its distinct, colorful bands of clouds in shades of white, red, orange, brown, and yellow. These immense belts move at different speeds, creating turbulence and powerful storms. The most famous of them – the Great Red Spot, a gigantic hurricane larger than Earth itself – left a lasting impression on me. It quickly became clear that I wanted to become a researcher. Even today, I still enjoy observing the night sky with my children, searching for the Perseids or tracing the constellation Cassiopeia – which also happens to be the name of my little daughter.

First, I moved to Brazil to complete my Master’s degree in theoretical and mathematical physics at the University of UNESP in São Paulo from 2007 to 2009. I then looked around to see where I could realize my dream of going deeper into astrophysics. And that’s how I ended up in Potsdam – at the renowned Leibniz Institute for Astrophysics. During my studies there, I discovered how programming could transform vast collections of astronomical data into models that reveal the hidden patterns and stories of the cosmos.

It is certainly an advantage to speak the language of the country in which you live and work. But for us in research, English is essential and our all-day language in the lab. As a data scientist, completely different languages suddenly play an important role, for example programming languages such as Python, SQL and R. Then there are various tools such as Power BI, Jira, Confluence, Slack, GitHub and MLOps.

It is often said that in our increasingly digitalized world, collecting and evaluating data is like digging for gold. I completely agree with this, and it is also true in skin science. Companies that are able to make important decisions based on a sound database have a clear competitive advantage. Data helps to identify correlations and trends and to optimize products and processes. AI systems are also absolutely dependent on being constantly trained with structured high-quality data.

Interesting positions in university research are unfortunately few and far between in the field of astrophysics. In 2015, I became aware that Beiersdorf wanted to fill a position as a Research Scientist in the Data Science AI/IoT Lab for the first time. I found this exciting, as skin care has always been important to me and NIVEA has also been present in Colombia since the late 1920s and is a very well-known brand. So I applied – and got the job.

I started out in the Biophysics department and development new data analysis approaches. This involved aggregating and analyzing relevant data from skin care studies conducted in the company over the past 20 years and establishing an appropriate basis for future AI models. Such models were also the focus of the skin hackathon we organized to look for new talents. The main award of the hackathons was a paid internship in our lab, which was a real win-win situation for us and the participants. One of the outcomes was a PhotoAgeClock, where we applied deep learning algorithms for the development of non-invasive visual biomarkers of aging based on Beiersdorf’s clinical photography data. I was also the project lead for the initial efforts to discover the opportunities to support formula optimization using machine learning. As part of that initiative, we identified the Uncountable platform, which remains our partner in this field to this day. In the summer of 2021, I seized the opportunity for the position of Senior Scientist in Data Science AI/IoT – and with it main parts of the data science responsibility for one of the world’s largest skin studies called SKINLY.

What is the Internet of Things (IoT)?

The IoT connects the physical world with the digital world. It enables “smart” devices to communicate not only with each other, but also with central applications and services. The networked “thing” often sends information to a cloud. There, the data is processed, made accessible or serves as the basis for further services.