Mohamed M.

Mohamed Mohudoom | Robotics Software Engineer I build the software that makes robots perceive, navigate, and act. From raw sensor data to autonomous behavior, I work across the full stack of a robotic system. Perception and 3D vision is where I'm strongest. I work with cameras, LiDAR, and depth sensors to build object detection, segmentation, point cloud processing, 3D mapping, and multi-sensor fusion (EKF/UKF) that holds up in real-world conditions, not just in simulation. I optimize these pipelines for real-time performance on embedded hardware like NVIDIA Jetson using CUDA. On the autonomy side, I build and tune navigation stacks in ROS2, including SLAM integration, path planning, and obstacle avoidance. I've worked with both differential drive and Ackermann robots, configuring everything from costmaps to trajectory planners like MPPI and Smac Hybrid A*. I also apply reinforcement learning and imitation learning to train robot policies for navigation and manipulation, with sim-to-real transfer on resource-constrained platforms. What sets me apart: I have a mechanical engineering background. I think about the full system, not just the code. Sensor mounting, kinematics, actuator constraints, URDF modeling, the things that determine whether your robot actually works or just looks good in Gazebo. I write clean C++ and Python, work primarily in ROS2, and can take a project from simulation to hardware deployment. Core Skills: C++, Python, ROS2, OpenCV, PCL, LiDAR, 3D Cameras, Camera Calibration, Point Cloud Processing, Sensor Fusion, SLAM, CUDA, Object Detection, Image Segmentation, NVIDIA Jetson, Autonomous Navigation, MPPI, Nav2, Reinforcement Learning, Gazebo, EKF/UKF, URDF