Perception Engineer

A Perception Engineer role in the autonomous-driving development office of a mobility startup working toward social implementation of self-driving taxi services with its own autonomous-driving software. The company is building development infrastructure — real-vehicle data collection, training-dataset creation, simulation, and model evaluation/improvement — and develops an end-to-end autonomous-driving model in which perception directly affects input quality. The role spans, cross-functionally, from sensor-data infrastructure design through recognition-model development, evaluation and improvement to raise overall perception-system quality: perception-algorithm development (research, implementation and improvement of object detection, segmentation and tracking; 3D spatial recognition and sensor fusion with camera and LiDAR); sensor-data infrastructure and representation design (calibration, coordinate transforms, time synchronization; data-processing pipelines; input representations for downstream models); and building data/evaluation loops (auto-labeling, hard-example mining, evaluation-data preparation, offline evaluation and real-vehicle log analysis). Based in Ota-ku, Tokyo. Full-time or contract.