Senior Robotics Software Engineer, Perception

Overview In this role you will develop real-time perception and sensor fusion for autonomous, battery-electric rail vehicles. You will work with cross-functional teams to bring multi-modal data streams from lidar, camera, IMU, and GPS into reliable state estimation and perception. You will translate first-principles methods into production-ready fusion and vision software. This position combines math, sensors, and field deployment to enable 24/7 autonomous operation and safer, greener freight solutions. ResponsibilitiesBuild and maintain real-time software to ingest and fuse high-throughput, multi-modal sensor data (lidar, camera, IMU, GPS).Implement, validate, and test real-time sensor fusion across simulation and real-world deployments, catching regressions early.Perform sensor calibration (intrinsic, extrinsic, temporal) and apply linear algebra, 3D geometry, and optimization to state estimation.Design and implement geometry-based computer vision and robotics algorithms for real-time perception; ML-based approaches are a bonus.Evaluate sensor selection and system-level tradeoffs and define test plans to prove 24/7 production reliability.Contribute to sensor system hardware design, including rugged enclosures and mounts for field use.Build and maintain sensor pipelines for real-time streaming and processing of sensor and video data.Collaborate with ML, firmware, embedded platform, and systems engineering teams to integrate perception into the autonomy stack. Key requirements4+ years of hands-on robotics or perception software experience in real-world systems.Strong background in geometry-based computer vision and robotics algorithms (multi-view geometry, filtering, optimization); ML experience a bonus.Proficiency in Rust and C/C++.Experience with real-time systems handling high-throughput, multi-modal sensor data and fusion across lidar, camera, IMU, and GPS.excellent communicationcollaborationability to work on interdisciplinary teamsgeometry-based computer visionmulti-view geometrystate estimation