We take you behind the scenes with Yves Daemen, project manager and team leader from advanced engineering at NavInfo Europe, to learn how he and his team bring our advanced AI technology into real-world solutions, and how they ensure our offering meets customers’ high standards.
Hi Yves, could you describe yourself and your role at NavInfo Europe?
I began working at Mapscape in 2007, when I was still a fresh graduate. I started working as a software engineer, and later became the project manager and team leader. My main role revolved around working on projects related to the creation of digital navigation maps for well-known European OEMs. In 2020, I transitioned to the Advanced Engineering (AE) department, where totally new domains were introduced to me and the team: AD Simulation and AI models.
My role is currently a mix of team leader and project manager: I make sure the whole team is running smoothly and that everyone feels good in his current role. On the other hand, I also guard the progress of the projects to ensure we are reaching the targets on time.
Can you tell us about your day-to-day activities?
One of my main daily tasks is to be the bridge between management and the engineering team. This means setting up meetings to facilitate smooth communication between the relevant departments to ensure that each team can meet their goals while being fully updated with each other’s activities. This way, the teams are synchronized with one another and can continue their work without too many disruptions.
Could you describe your team?
My team is a great mix of experience and technical expertise. This combination provides an ideal setup to produce high-quality solutions for our customers, as well as quick turnaround times for POC developments of research.
Can you discuss your experience adjusting your expertise to work on advanced AI engineering?
One challenge I faced was moving from working on engineering in the Data delivery Services team to working on AI engineering projects. This was indeed quite a challenge due to many reasons, for example, the projects in DDS have to adhere to very high-quality standards like ISO and A-Spice, and the whole software engineering environment is built around those principles. On the other hand, AI projects tend to be diverse and require more efforts on the development side. We had to set up our own processes and practices, as well as setting up solutions to ensure the stability and reproducibility of the AI-based solutions. What helped me was that I could use the experience from the more strict approach in Data delivery Services and apply the most valuable practices to our team’s work.
What is an interesting project that you recently worked on?
One of the most challenging, yet exciting projects was the video anonymization service that we have running on AWS. This project combines several technical challenges, like building a system on AWS which uses a mix of AWS-managed services as well as a highly scalable Kubernetes cluster for the processing capacity. With this cluster, we can scale up to hundreds of GPU-enabled servers on AWS. On a monthly basis, we process close to 500 TB of video data. Besides providing such a scalable, fully automatic system, we also overcame the challenge of tuning the model for optimal cost efficiency.
Following the last question, how would you describe our GDPR compliance solution?
I believe that many companies are still in the stage of understanding the scope of GDPR and are actively looking for solutions in that area. For most companies, it’s not their core business, so this is where we come in and take care of an end-to-end solution for the customer, with an extremely simple interface. It takes away all the worries for a customer in the area of GDPR. So far, we have built a system for high-volume video processing and a real-time image anonymization system which can also be tried out on the company website. We plan to extend our standard offering in the future, as to have a very cost-efficient solution available for any type of customer that is in need of making their video or image data GDPR compliant.
What efforts have been done by you and your team to make this offering more competitive?
Our major focus points are 1) the ease of integration for customers. We aim to provide easy interfaces to our system, but we are open to suggestions from our customers, and we can always tailor the model for their specific needs. 2) making our solutions scalable yet cost effective – this is already included in the design phase.
What is next for your team in terms of integrating advanced AI research into real-world applications/smart mobility?
Recently we have been doing some investigation into combining our expertise in AI-based computer vision solutions with our navigation experience. This has resulted in 2 interesting projects: based on panoramic imagery, we have built a fully automatic software pipeline for 1) the creation of a traffic sign map and 2) creation of a polygon map that indicates the type of ground cover. Both functionalities could potentially help to reduce a lot of manual work that is done nowadays.
