Can you tell us about your job?

Quentin Grimonprez – The heart of my job is to design and adapt analysis methods to our customers’ constraints. On the one hand, I work with customer data, i.e. our customers send us their data and we carry out analyses combining statistics and Machine Learning to achieve their specific objectives (consumption optimization, prediction…).

On the other hand, we develop and industrialize in-house our own algorithms (segmentation, clustering, etc.) which make up our standard Pipeline used in our real-time monitoring application. Of course, these algorithms need to be monitored, maintained and developed. There’s also a whole R&D part to it: bibliography and reading articles to improve on a daily basis and design new methods.

Alongside all this, I teach advanced modeling courses at Polytech’lille. The aim is to teach techniques for tackling recurring data problems: what to do when data is missing? What methods can be applied to large-scale correlation problems? etc.

What does Head of Data Science mean in everyday life?

Q.G – On a day-to-day basis, I liaise between the data scientists and the developers regarding the algorithms integrated into our app. I represent the data team at developer meetings to highlight our needs, and I bring up any changes that will impact us.

On the customer side, at the start of a project I take part in the initial technical meetings. Once the customer’s objective is well defined, I assign projects to the data team according to availability and skills/preferences.

You’ve been with us since the start of the DiagRAMS adventure. How did you make the transition from research to start-up?

Q.G – I started my career at Inria ( the French national digital science research institute) in the MODAL team, a research team specialized in statistics and machine learning. In fact, I did my thesis there on time series applied to the health sector, and more specifically on the study of the genome of cancer sufferers. The whole aim was to find the genes linked to disease relapse, based on a study of very few individuals.

I then met Margot and Jean-François, who were working on technology transfer at Inria, and joined them in the DiagRAMS adventure as an associate. Today, I’m transposing the methods I studied in my thesis to the industrial sector, where there are few breakdowns in learning datasets.

What are your favorite challenges in your job?

Q.G – What I like best is research applied to real-life problems. But also, learning new things: searching, discovering new methods, understanding them… That’s why I’ve launched in-house data science presentations at DiagRAMS. The aim is to discover methods and packages and present them in turn, so that we can all learn together. 

Bonus question: what’s your incredible talent?

Q.G – I’m 3rd nationally in Quizzroom.