Spent a decade at Eurailscout Inspection & Analysis building end-to-end deep learning systems for national-scale railway infrastructure inspection — defect detection & segmentation pipelines (YOLOv8, InternImage, TensorRT, CUDA), multi-sensor data frameworks across optical, ultrasound, and eddy-current modalities, real-time video analysis on inspection trains, annotation tooling (PyQt5), REST APIs, and cloud ML experimentation (Azure ML, MLflow). Published cross-sensor fusion research at ECNDT 2018 (first author) and WCNDT 2016.
M.Sc. at TU Munich in collaboration with SCOPIS GmbH — built an online camera self-calibration system for surgical endoscopes using a Gaussian Sum Filter within a monocular SLAM pipeline, enabling real-time intrinsic estimation without a calibration target.
Now independently building agentic LLM systems and cloud-native data pipelines — and actively looking for the next challenge as an AI Engineer or Data Engineer in Berlin.
