A scientist in industry: meet Arthur HouwelingPublished on: Author: Team Communication Category: News
In 2016 Arthur Houweling moved from academia to industry. As team lead data science at Qualogy, he has benefited greatly from his scientific background. Meet Arthur and discover what he and his team can do for organizations.
“I joined Qualogy as team lead data science in early 2016. I work with my team on projects for various private and public sector organizations. As a product owner or analytics translator I help identify, develop and implement AI use cases. I bridge the gap between the business and our data scientists. During development, I continuously monitor whether our solution is of value to the customer.”
“As a data science team lead, I have benefited greatly from my scientific background. After my studies in Computer Science and Biology, I did my PhD at the University of California, San Diego. That was the ideal place for my thesis research into neural networks. The region is a vibrant hub of academic and industrial groups in machine learning and AI.”
From academia to industry
“After my PhD and postdoctoral research in Rotterdam and Berlin, I joined Erasmus University as an assistant professor and led my own research group. Life as a scientist is fascinating, but it also had its limitations for me. For example, the applications of my fundamental research were not always immediately clear. Furthermore, I have a broad interest, while as a scientist you typically focus on strongly related issues during your career. I regularly considered switching careers. Early 2016 the time had come.”
Experience in industry
“At Qualogy I have gained a lot of experience in initiating, defining and supervising data science projects for our customers. The topics of our projects vary enormously: from machine learning, deep learning and data analysis to natural language processing, computer vision and operations research. This makes my work challenging and interesting.”
Connection with data scientists
“I enjoy working with data scientists. They are passionate and inventive people. I like to analyze problems and formulate solutions together. We do this in weekly sprints. In consultation with the team, I define the stories and prioritize the backlog.
When working on a story, a data scientist sometimes opts for an unnecessarily complicated solution. It is up to me and the team to indicate how things can be simplified. There is nothing like writing code yourself again.”
Critical assessment of outcomes
“It sometimes happens that the results of an analysis or the predictions of a model are impossibly correct. Dealing with this is also part of the job of a data scientist. By making testing and peer review an intrinsic part of the work process, the team learns to critically evaluate its own results. This is how we ensure to deliver quality.”
A selection of our projects
Arthur is happy to tell you more about the versatile projects of Qualogy's data science team. Click here for more information about the team and our services.
Distribution center optimization
“I worked with my team on a challenging optimization problem for the distribution centers of a large supermarket chain. After visiting the distribution centers and talking to IT specialists, I mapped out all relevant logistics processes and algorithms.
During a proof of concept, we then used advanced algorithms to reduce the number of packages and collection time of online orders. The results convinced management to replace the algorithms and thereby reduce operational costs in the logistics chain.”
Real-time football analytics
“My work is very appealing because of the great variety in projects. For example, we are developing a football coaching application based on real-time player tracking data obtained during matches and training sessions. The system generates tactical alerts on the coach’s smartphone or tablet. The application is still under development, but a prototype has already been successfully tested by the Dutch national team.
The idea for the application arose after a meeting at the KNVB with data analysts from various Eredivisie clubs. I joined the meeting to present the results of our data analyses during the 2017 UEFA Women's Championship. After the meeting, I made an inventory of the wishes for real-time analytics and translated these into product requirements. We hope to be able to present the application soon.”
Ideal mix of tenants in shopping centers
“Sometimes customer questions are extremely challenging, requiring an original approach. IKEA Centers, for example, wanted to use predictive analytics to determine the ideal mix of tenants in new shopping centers they are developing around the world. By linking internal and external location data to store performance indicators, the data science team has developed a machine learning model. We use the model to predict the sales figures and visitor numbers of prospective tenants for a location, based on the proximity to other stores and location features.”