Computer Vision ML / Research Engineer

Mission: Own the perception engine that turns stereo camera captures into accurate, auditable pile measurements.

JD: We are building a computer vision system for industrial inventory measurement using stereo cameras, depth estimation, segmentation, point clouds, calibration data, and 3D volume estimation. You will own the core CV/ML pipeline: stereo reconstruction, segmentation, point-cloud cleanup, ground-plane estimation, stitching, volume calculation, and model evaluation. This role sits at the heart of the product.

You will own: FoundationStereo/s2m2/YOLO-style model integration, 3D reconstruction quality, volume accuracy, calibration assumptions, failure-mode analysis, confidence metrics, and the path from research pipeline to production-grade measurement.

Must have: Strong Python, PyTorch, OpenCV/Open3D, 3D geometry, camera calibration, stereo/depth, point clouds, segmentation, and comfort debugging real-world sensor data.

Nice to have: CUDA, robotics perception, photogrammetry, SLAM, industrial inspection, geospatial/volumetric measurement, Modal or GPU cloud experience.