Germany
Computer vision researcher specializing in deep learning and vision-language models (VLMs). Passionate about solving complex AI challenges with a bias to action. Strong problem solver with a proven track record of building AI pipelines, leading development teams, and delivering high-impact solutions. Excellent presentation skills with experience engaging large audiences at AI conferences and industry events. During my PhD period, I researched AI methods for change detection in HD maps for automated vehicles in LiDAR point clouds. Further, I created the large-scale HD map dataset 3DHD CityScenes with an accompanying deep leaning pipeline for object and change detection. Thesis: https://amzn.eu/d/04Hauox5 Dataset: https://zenodo.org/record/7085090 DevKit: https://github.com/volkswagen/3dhd_devkit My research project was the first to be published on GitHub in the name of Volkswagen after approval by executive board members.
Department lead for automated driving (AD) engineering, full-stack from perception to driving function. Designing stack architecture, AI & data frameworks, and managing multiple teams to success. AD made in Germany.
Designing AI-powered data pipelines enabling ODD management, scenario extraction, and AV testing. Architect of a data-driven toolchain for designing road networks w.r.t. AV testing and operations. Design of a VLM-based data fusion pipeline that extracts ODD-relevant data from images for subsequent road segment annotation, enabling road-specific ODD management. Design of data pipelines to extract environmental data (e.g., rain, snow, visibility) from crowd-sourced fleet data and images using VLMs. In parallel, working as technical program manager w.r.t. ADMT's map program, including concept assessment, risk identification, and roadmap management.
3D map evaluation in LiDAR point clouds using deep neural networks. Research of deep learning-based methods for change detection in HD maps. Development of a machine learning pipeline from scratch. Extensive experience gain in Python and PyTorch.
Automatic dataset annotation. Developing tools for automatic annotation of datasets, e.g., intelligent editing tools or AI-based anonymization. Development of camera calibration methods for research vehicles. Creation of software modules for real-time dataset recording using research vehicles.
Fleet-based generation of weather maps. Method for fleet-based measurement of rainfall: https://patentimages.storage.googleapis.com/0f/f3/6d/505bf185586170/DE102015209601A1.pdf Fleet-based environmental measurements using Kalman filter: https://patentimages.storage.googleapis.com/b2/51/63/c44a2d751e83d2/DE102015209602B4.pdf
Focus: e-mobility. In-factory assignments, e.g., assembly of machine controls with logic control units.