Vision & Perception Engineer for Humanoid Robots (Human)

As a Robotics Software Engineer focused on vision and perception, you will be the architect of our robot's "eyes" and its ability to comprehend the world. You will be responsible for designing, developing, and deploying the complete perception stack that enables our humanoid robots to see, understand, and interact with complex, dynamic human environments. Your work will transform raw sensor data from cameras, LiDAR, and IMUs, Force Torque and joint sensors into actionable, high-level understanding for navigation, manipulation, and human-robot interaction. You will work extensively with 3D computer vision, deep learning, sensor fusion, and high-performance C++ to build a robust and reliable perception system.

• Design and implement the complete perception software architecture, from low-level sensor drivers to high-level scene understanding, ensuring real-time performance and robustness.
• Develop and optimize state-of-the-art algorithms for SLAM (Simultaneous Localization and Mapping), object detection and tracking, semantic segmentation, and 3D reconstruction.
• Write clean, efficient, and maintainable code in modern C++ (C++17/20) for the core perception and sensor fusion libraries.
• Architect and manage the high-throughput flow of sensor data through the system using ROS2 and its underlying DDS framework.
• Integrate and calibrate a diverse array of perception sensors, including stereo/depth cameras, LiDARs, and IMUs, into a unified world model.
• Collaborate closely with controls, manipulation, and planning engineers to define perception requirements and deliver the necessary data for downstream tasks.
• Develop tools for data visualization, system diagnostics, and performance evaluation of perception algorithms.
• Create and maintain comprehensive technical documentation for the perception system, sensor interfaces, and data formats.
• Validate perception modules both in simulation and on real humanoid platforms, bridging sim to reality.
• Ensure perception modules meet real-time and safety-critical requirements for human-robot collaboration.
• Establish and champion best practices for software development, including version control (Git), continuous integration (CI/CD), and data-driven automated testing.

• Education: Bachelor's or Master's degree in Computer Science, Software Engineering, Robotics, or a related technical field.
• Professional Experience: 3+ years of professional experience in software development with a focus on computer vision or perception for robotics.
• C++ Proficiency: Strong command of modern C++ (C++11 and newer), including object-oriented design, multithreading, and performance optimization on memory-constrained systems.
• Computer Vision Fundamentals: Solid theoretical and practical understanding of 3D geometry, camera models, multi-view geometry, filtering (e.g., Kalman filters), and sensor fusion techniques.
• 3D Vision Experience: Demonstrable, hands-on experience developing and deploying algorithms using 3D data, particularly with point clouds. Proficiency with libraries like PCL (Point Cloud Library) is essential.
• Deep Learning Expertise: Proven experience in applying deep learning models for perception tasks (e.g., object detection, segmentation). Familiarity with frameworks like PyTorch or TensorFlow and experience deploying models on robotics hardware.
• Experience with GPU programming for accelerating perception algorithms (e.g., CUDA, OpenCL).
• Experience with performance optimization for real-time applications on embedded systems (e.g., NVIDIA Jetson).
• Familiarity with simulation environments such as Gazebo or NVIDIA Isaac Sim for perception development and testing.
• Proficiency in Python for rapid prototyping, data analysis, and training machine learning models.
• Experience with MLOps and infrastructure for managing large-scale sensor datasets and training pipelines.
• Knowledge of classical computer vision libraries, such as OpenCV.