Our client is revolutionizing autonomous driving with a unique approach rooted in cognitive neuroscience and cutting-edge German research. Their mission is to make intelligent, explainable decisions in complex traffic environments without relying on massive datasets. As one of the first companies in Germany to pursue level 4 certification for autonomous vehicles, they'refocused on safety, transparency, and scalability to shape the future of mobility—connecting people wherever they are.
As a Geospatial Segmentation AI Engineer, you’ll develop the deep learning systems that enable our vehicles to perceive and understand the world through satellite, aerial, and sensor-based imagery.
Responsibilities:
• Design and implement AI models for semantic and instance segmentation using satellite, drone, and LiDAR imagery.
• Develop preprocessing pipelines for geospatial image data using tools like GDAL, Rasterio, and GeoPandas.
• Train, evaluate, and fine-tune CNN architectures (U-Net, Mask R-CNN, DeepLab, etc.) for high-resolution remote sensing tasks.
• Integrate segmentation outputs into perception systems for urban/rural mapping, land use analysis, and road understanding.
• Work with large-scale geospatial datasets (GeoTIFF, NetCDF, shapefiles, GeoJSON) and manage cloud-based geoprocessing workflows.
• Build automated data pipelines for labeling, training, and validating geospatial models using tools such as CVAT, Labelbox, and QGIS.
• Collaborate with perception, software, and sensor teams to align map data, real-time vision outputs, and spatial AI models.
Requirements:
• Strong experience with geospatial data processing and GIS tools (e.g., QGIS, GDAL, GeoPandas); understanding of CRS (e.g., WGS84, UTM).
• Hands-on expertise in deep learning for image segmentation using frameworks like PyTorch, TensorFlow, or Keras.
• Experience with satellite/aerial image analysis and handling raster/vector data formats.
• Familiarity with CNN architectures like U-Net, HRNet, DeepLab, and object detection methods (YOLO, Mask R-CNN, etc.).
• Proficient in Python and libraries such as OpenCV, scikit-image, Rasterio; experience with Jupyter, Docker, Git.
• Knowledge of cloud platforms and services for geospatial processing (e.g., AWS SageMaker, Google Earth Engine) is a plus (negotiable).
• Experience with GPU-based training, distributed learning, and handling largescale Earth observation data.
Why You Should Join Us:
• Work in an intellectually stimulating and innovative environment where you can take full ownership of your projects.
• Enjoy flat hierarchies, an open culture, and fast decision-making processes.
• Collaborate with a skilled and dedicated team eager to share their knowledge and expertise.
• Be part of a multinational workplace that values diversity and integrates different backgrounds and perspectives.
• Work in the vibrant heart of Berlin, in the dynamic Kreuzberg district
