Choose your country / language

Under the Digital Hood: Cracking the Secret Code of Data Science

Welcome to our very first edition of “Under the Digital Hood” Series. Digitalization is one of the hottest topics in the industry world right now, and it comes with a bunch of funky buzz words and job roles that never existed before. Think data science, think internet of things, machine learning, AI, digital twins, big data, deep learning and the list can go on and on. But, what do all of these actually mean? Well, let’s hear it (read it, in this case) from the people who are behind the digital transformation. In the upcoming editions of our series, we plan to go under the digital hood and find out more about the driving forces powering the digital world. And who could be a better fit for this than our very own Oerlikon Digital Hub team members.

Under the Digital Hood: Cracking the Secret Code of Data Science

We start with “the sexiest jobs of the 21st century” as Harvard proclaimed it: the Data Scientist. We sat down with Dr. Pit Vanhoefer, Senior Data Scientist in our Digital Hub, and worked on cracking the secret code of his role. Enjoy the interview and let us know your thoughts!

Digital Hub: So, let’s get the party started! Pit, tell us a bit about yourself and this role of yours that has everybody talking ?

Pit Vanhoefer: Hi! I started as a Senior Data Scientist in the Digital Hub in September this year. My background is in High Energy Physics, where I was trained in data analysis and big data, studying antimatter produced at particle colliders. I then moved into the industry world, and worked for a couple of years as a Data Science Consultant, mainly for the Automotive industry.

In my current role as a data scientist, I am responsible for developing machine learning models to improve our industrial products. A common use case for this is predictive maintenance. Known failures can be used to label (e.g. OK / failure) machine data that we pick up from sensors. This labeled data is then passed to a computer that will define rules (or “train a model”, as we say it), that can then be used to predict an upcoming failure when looking at new data. Hence, we can prevent the failure if we (or even the machines) know how to react.

DH: Improving industrial products must be quite an exciting task to be working on. Can you describe a bit further the data scientist role, what it consists of?

PV: My role is quite diverse and goes from hands-on coding to project management. That means I contribute to setting up new projects with a clear focus on data analytics, sharing my experience with colleagues and developing solutions – from the white board to a solid product. Another important aspect is to advertise a new mindset. A data driven company should make decisions based on insights derived from a solid and careful data analysis, instead of letting the highest paid person decide based on his / hers gut’s feeling. To drive this spirit is also part of my role. In fact, machine learning itself is a change of paradigm compared to classical programming.

DH: Wow, that seems like a very complex role! What are the actual skills that one needs in order to nail down all these different aspects of the job?

PV: Besides mathematics, data management skills and a little programming, I think a fast understanding of the situation, and the ability to map the available information to the business case is certainly helpful. When data analytics enters more unknown regimes, experience becomes quite handy. There are many exciting methods out there and one can easily get distracted. So, key is to focus on the business case.

DH: So, how does your day actually looks like? What do you do after you have your kick starter coffee, what are your responsibilities?

PV: Since my team and I often have to align our path and adopt to freshly gained insights, fruitful conversations are daily activities, often followed by productive coding and designing. We try to reduce the number of meetings by working closely together with the relevant people and document our work in a digital way, accessible to all team members. Finding and testing optimal solutions adds also project management tasks to our work at the Digital Hub.

DH: Let’s get a bit deeper into your process. What are your steps, how do you tackle tasks, what are some of the challenges you face?

PV: Data science is usually not a linear process, as there is no standard recipe that can guarantee a certain outcome. Usually starting with a proof of concept is a good idea. Having more potential projects in the pipeline and using a fail-fast approach allows to focus on the promising ones, after a first evaluation and identifying necessary actions, for others. The Cross Industry Standard Process for Data Mining (CRISP), as depicted below, is a common process for bringing analytics project to success in an iterative manner.

In the industry world, machine data is usually generated with the purpose of building and operating the machines, and not for deriving a model for prediction. Therefore, it highly differs from case to case how useful the data actually is. Common challenges are dirty data, missing labels, insufficient amounts of or high variety in the data.

DH: What are the goals and deliverables of your role?

PV: Ultimately to enhance our product with smart algorithms and to increase efficiency by reducing repetitive manual work. Since especially the former is a long term goal, defining this path is also part of my job. Not every idea becomes a product, so an honest evaluation at an early stage, together with the necessary steps to reach the required level, can also be a deliverable. In the best case, we deliver production ready solutions which make use of machine learning or artificial intelligence.

DH: And how do these goals relate to business goals? As I am sure, some of the times, these two worlds might not seem to be connected.

PV: Well, for starters, for our own company, the benefits of data science will be seen in the product portfolio. Besides the above mentioned predictive maintenance capabilities, data analysis can also be used to improve the quality of our products through a better understanding of the underlying processes. The list of use-cases that make use of data analytics is long. Furthermore, these capabilities can open new horizons for Oerlikon, enabling the company to share its solutions in the future and, thereby, entering complete new business opportunities.

DH: By now, I hope all our readers are aware of the digital transformation that Oerlikon has embarked on, with the Digital Hub as the trailblazer. So, my next question is: how does your role as a data scientist contribute to the digital journey of our company?

PV: The Digital Hub team has the opportunity to contribute to, and shape many of the challenging and exciting aspects of this digital transformation. Besides setting up a trusted and visible data science team, we also need to build an environment that allows to fully make use of our data’s potential. Here, the entire company needs to be on-board and share the same vision of the path into a digital enhanced future. And I hope to serve as an example for this purpose.

DH: Pit, we have reached the end of our interview. Thank you very much for being part of our “Under the digital hood” series and for answering these burning questions we all had about your super role.  Before we wrap it up, one finally thing: can you tell us one misconception about your role?

PV: I am often confronted with the misconception that good results can be achieved in almost no time. Although one can set up a good looking PoC quite fast, reaching production readiness can take some time and effort, in my experience.

DH: Well, thank you for spending your time with us today, Pit!

There you go, folks. A real life data scientist (he might seem virtual, but you will just have to trust us), right before your very own eyes. As we continue with our series, you will find out more about Data and the people that work with it every day, about the Internet of Things, e-commerce, digital projects, and the new ways of marketing used in this digital era.

If you have questions that require an answer in these areas, feel free to send them to us via our Linkedin channel. And do let us know your thoughts and comments about this edition, we would love to have a conversation.

Until the next time, may the digital force be with you all!

Contact

Oliver Wachsmuth

Oliver Wachsmuth

Global Head of Oerlikon Digital Hub

Feedback or questions? Get in touch with us!

keyboard_arrow_up