Senior Vision & Perception Engineer

CDDS is a company focused on developing next-generation control, guidance, and state-estimation algorithms for high-speed kinetic interceptor drones. The Senior Vision & Perception Engineer will be responsible for architecting and optimizing the low-latency vision pipeline on NVIDIA Jetson platforms. This role involves developing, tuning, and extending DeepStream-based pipelines in C++, integrating differential-imaging workflows, and ensuring real-time performance. The main day-to-day responsibilities include implementing and optimizing vision pipelines for object detection and tracking, developing high-speed detection and tracking algorithms, optimizing TensorRT inference performance, profiling and tuning performance, and maintaining GPU-accelerated development containers and CI pipelines. The engineer will collaborate closely with robotics, control, and embedded engineers to build the perception backbone that enables high-speed interception. To excel in this role, candidates must have an MSc/PhD in Computer Vision, Robotics, Computer Science, or a similar field. Expert knowledge in DeepStream, GStreamer, and low-level video pipelines is essential, along with strong C++ skills and experience with CUDA, TensorRT, or Jetson optimization. A strong background in modern detection models such as YOLO and EfficientDet is also required. Experience with ROS2 and multi-process robotic systems is a must. Nice-to-have skills include comfort with rapid field-testing cycles and high-ownership environments. The role is based in Schlieren (Zürich) and requires on-site presence. The team at CDDS is highly technical, ambitious, and hands-on, working in a fast-moving startup environment with real-world testing and deployments. What the role asks for: - MSc/PhD in Computer Vision, Robotics, CS, or similar - Expert knowledge in DeepStream, GStreamer, and low-level video pipelines - Strong C++ skills and experience with CUDA, TensorRT, or Jetson optimisation - Strong background in modern detection models (YOLO, EfficientDet, etc.) - Experience with ROS2 and multi-process robotic systems