A portrait of five data scientists
Talented linguists, gold miners, business experts, researchers of the future – Swisscom’s data scientists are real all-rounders. Their personal qualities make their work unique.
Text: Christoph Widmer, Images: Philipp Zinniker, December 2017
They are called Andreea, Mohamed, Gaël, Sergio and Florian. They are from five different countries and have quite different professional backgrounds. What they share is a passion for data and the analysis of it, and they call themselves data scientists. We will introduce you to five personalities from different units within Swisscom and show you how their character traits and interests influence their work.
Andreea Hossmann has a real gift for language. In addition to her mother tongue, the Romanian native can also speak French, English, Spanish, Italian – and fluent German. “Communication skills are often not a very strong trait of developers,” she says. “But it is one of my strengths. After two years and a period of maternity leave, I now manage a team of ten working in the area of Artificial Intelligence – not least thanks to my encouraging and flexible supervisor as well as to the full support of my family.”
As she did her PhD at ETH Zurich, Andreea Hossmann was able to add German to her list of foreign languages.
Hossmann’s team developed, among other things, a system for the Customer Support department that analyses customer queries and automatically assigns them to the appropriate specialist as well as a tool that allows contracts between Swisscom and its customers to be categorised. In the future, they intend to make the interaction with these systems more natural by using speech recognition. Hossmann doesn’t just bring her knowledge of foreign languages into the mix. She earned her degree in telecommunications engineering in France – and added on a Master of Research and a year abroad at the UIUC in Illinois. She received her doctorate from ETH Zurich – this was due to her interest in languages: “The options I was considering were Cambridge, Oxford or ETH Zurich”, she says. “I chose ETH Zurich because I also wanted to learn German.” Hossmann now lives with her husband and two children in Bern.
Her research background is very often a great asset also at Swisscom – either through her work on data analysis or in the collaboration with the EPFL in Lausanne, where her team is developing new AI topics together with EPFL professors and students. Her programming does, however, tend to fall by the wayside. But Hossmann knows how to handle that: “I work from home one day a week. I make sure that there are no upcoming meetings, so that I can concentrate on programming.” In this way, she also manages to keep good command of her programming languages.
For Mohamed Kafsi, human mobility wasn’t just the topic of his PhD; he has also been investigating it for the last two years at Swisscom: Kafsi works as a data scientist at Swisscom Big Data and Mobility. He is currently using mobile phone connectivity data to reconstruct and predict the flows of people’s movement. “We help cities to better understand mobility,” he says. “With our City Insights solution, we can continuously provide informative movement data to urban planners, for instance.”
Thanks to his internships at Deutsche Telekom and Nokia, Mohamed Kafsi was able to gain telecommunications knowledge during his degree.
The main thing Kafsi needs to do for his work is to dig. “From my point of view, data scientists cannot shy away from a hands-on approach, including rummaging through the data,” he says. “Around 80 percent of my work is in data preparation. Only when the data has been modelled can I start to analyse it and find insights.” To this end, Kafsi also acquired telecommunication-specific knowledge: When he was 19, he left his homeland of Tunisia to study Communication Systems at the EPFL in Lausanne – and completed internships at Deutsche Telekom and Nokia.
He also deems it essential to have a good dollop of curiosity. It is what motivates Kafsi to try out new sports, for example, to browse through books or to peer through his camera viewfinder when taking photos – and he is happy to get a break from data when at home. “Luckily my wife isn’t a data scientist, but rather a lawyer,” laughs the new father.
As a data scientist on the Big Data Network Intelligence Squad, Gaël Grosch helps the network engineers to gain an overview of all incidents that occur in the Swisscom network. “We collect network data and try to detect failures,” he explains. “Thanks to the various types of data, we can trace the events that occur during an entire phone call and thus better reveal why calls are sometimes dropped.” Grosch and his team developed the analysis platform Nicer for this purpose: it gives users an overview of network events such as calls, text messages and data usage – in near real-time.
Gaël Grosch considers Switzerland to be very advanced technologically, but he says that companies still need to learn how to use data science effectively.
At the heart of Grosch’s work lies a single principle: the benefit to Swisscom. To do his work, he needs a deep understanding of the business. That is the only way, he says, to really be able to implement business-optimising solutions – instead of just generating some pretty statistics. According to him, that is one area where Switzerland needs to catch up: “Business-oriented US companies such as Palantir are able to convey an ideal of data science,” he explains. “But an understanding of how Big Data can really be put to use for a business is in fact still quite rare in this country. Many Swiss companies need to first learn how to use data science sensibly and effectively.”
From a technological point of view, however, he says that Switzerland is already very advanced. Grosch particularly noticed this during his Communication Systems degree at EPFL in Lausanne, when he completed a semester abroad in Colombia. “Their entire infrastructure was slower than ours, the Internet didn’t always work and it took a long time to find someone who could help you. In Switzerland, on the other hand, you can always work efficiently, anywhere and at any time.”
A knowledge of computer science, mathematics, statistics, abstract thinking, analytical abilities, business know-how and an adeptness in new technologies: for Sergio Jimenez, those are the basic skills of a data scientist. Physicists are one group of people who can demonstrate these abilities. “That is why physicists are in such demand as data scientists,” he says. “Many of my former PhD colleagues now work in data science.”
For Sergio Jimenez Otero, data scientists and physicists share certain core skills.
The native Spaniard himself studied general, experimental and theoretical physics at the University of Glasgow. After he obtained his PhD in experimental physics at the EPFL in Lausanne (on the topic: “Why is there more matter than antimatter?”) and also did research on this topic at CERN, Jimenez then became a consultant at Swisscom. He helps internal and external customers to optimise their business processes – by drawing up business cases, creating proofs of concept and developing tailor-made solutions for the customer. The solutions include predictive analytics and the classification of documents: “One of our tasks is to help companies gain an overview of their text data,” he explains. “That allows them to find, for instance, legal documents faster and to examine them in a more structured manner.
In his free time, Jimenez doesn’t just enjoy watching TV series; biking and swimming are also among his hobbies – and for a good reason: “I find sport particularly important as it also promotes dynamism and creativity at work.”
As a data scientist within the area of the Industrial Internet of Things, Florian Pitschi develops solutions for predictive maintenance, among others for the air-conditioning company Walter Meier. “Thanks to sensor data, service usage data and machine data, we can detect which heat pumps have a high probability of failing soon,” he explains. “With our solution, the service technicians at Walter Meier can detect and remedy problems ahead of time.”
As a data scientist, Florian Pitschi can use data to create real added value.
As a “machine oracle”, Pitschi extracts real benefits out of data. And that’s what he always wanted to be able to do. After his degree in computer science and physics in Ulm and his doctorate in China, however, he first worked as a business intelligence consultant – but he found that it didn’t satisfy him: “It was never enough for me,” he says. “You display the data, perhaps you calculate a few trends, but you don’t learn very much from it. I wanted – and still want – to use data to create real added value, to optimise processes and to predict the future.”
Pitschi also trains these skills by taking part in events such as machine learning competitions: “I once even had to teach a skeleton how to learn to walk by itself, and I’ve used passenger data to predict which Titanic passengers would survive the disaster and which wouldn’t.” Pitschi is also an enthusiastic beach volleyball player – and that’s without knowing the final result. “That’s the exciting thing about it: in sport, sometimes the underdog wins.”
More on the topic