ML разработчик

LiveEO provides satellite analytics that model risks to customers’ assets and infrastructure from vegetation, ground deformation, and change detection.

• Drive geometric computer vision development for stereo and multi-view 3D reconstruction models and image matching and registration pipelines;
• Research and adapt state-of-the-art approaches in 3D reconstruction, depth estimation, feature matching, and adjacent geometric computer vision;
• Improve the generalization of learned stereo across sensors, geographies, and acquisition conditions using synthetic data and sim2real transfer strategies;
• Contribute to segmentation, detection, and change analysis projects;
• Own standardization, preprocessing, and quality diagnostics for high-resolution Earth observation imagery;
• Build scalable training and evaluation pipelines across cloud and secure on-premises environments;
• Deliver production-ready inference interfaces, model packaging, deterministic evaluation, monitoring, and constrained-compute adaptations;
• Collaborate with the data annotation function on labeling guidelines and edge cases;
• Work with partner teams to turn model capabilities into validated deliverables;
• Collaborate with external researchers and present findings clearly and efficiently.

• Strong computer vision fundamentals, including representation learning, supervision strategies, and evaluation design;
• Practical experience with stereo or multi-view reconstruction, depth estimation, or image matching and registration;
• Strong Python engineering fundamentals and experience writing clean, maintainable code;
• Deep experience with PyTorch and implementing and training deep learning models at scale;
• Strong understanding of ML experimentation, versioning, and tracking;
• Background in remote sensing, computer science, physics, a related field, or equivalent practical experience;
• Comfortable working with researchers and presenting findings clearly and efficiently;
• Eligibility to obtain a German security clearance (Sicherheitsüberprüfung);
• Ownership mindset and proactive approach to moving work forward;
• Clear communication and smooth collaboration within and across teams;
• Pragmatic approach to balancing deep research with practical delivery;
• Ability to work with complexity and turn ambiguity into structure;
• Nice to have: PhD in a relevant field, hands-on satellite or remote-sensing imagery experience, synthetic data generation, sim2real or domain adaptation for geometric vision, structure-from-motion, SLAM or visual odometry, neural 3D representations, NASA Ames Stereo Pipeline, MicMac, COLMAP, DSM generation, constrained-compute or edge-device deployment, model compression, quantization, optimization, Ray, Prefect or similar workflow orchestration, AWS, secure on-premises or HPC environments, SLURM, Docker, DVC, GDAL, Rasterio, GeoPandas, STAC, PostgreSQL or similar, SAR with optical imagery, geospatial foundation models, VLMs, self-supervised learning, contrastive learning, masked modeling.

• Flexible working hours;
• Overtime is offset with time off and rest;
• Collaborative learning environment with internal workshops, knowledge-sharing sessions, journal clubs, and hackathons;
• Office in central Berlin Kreuzberg with free fruit, nuts, and drinks;
• Potential participation in the employee stock option program;
• Urban Sports membership and BVG subsidy;
• Corporate pension program;
• Diverse international environment with more than 30 nationalities.