Analyzes fleet-scale video and sensor data (RGB, LiDAR, telemetry, bounding boxes) from robotics and AV teams into structured behaviors and validated events. Produces training-ready datasets and accelerates manual review by 100x for autonomy development.