Currently working on improving LiDAR 3D object detection algorithms

ADASTEC provides SAE Level-4 Autonomous public transportation solutions and we have deployed products on countries such as US and Norway• Made our pointcloud registration system faster by reducing HD map loading time by 37%, by implementing parallel computing features(OpenMP) for our voxel grid filter algorithm.• Developed utility product to help our deployment teams to monitor the performance of our mapping systems in real time.• Researched and unit tested various pointcloud registration algorithms for SLAM&Localization applications in order to improve ours.• Worked on 3D point cloud generation system

• Improved our testing pipeline of vehicle localization systems by developing test automation tool.• Integrated test automation processes into Unreal Engine environment for real time simulation testing.Used C++, Python, QT, ROS, Bash, Unreal Engine (CARLA)

• Developed various 2D video games for customers. Used Java, Python, C++• Implemented a real time camera calibration tool. Used C++, QT, OpenCV• Created Reinforcement Learning powered AI agent to play custom games. Used Python, Pytorch• Developed various task automation tools to helped our customers. Used Python

Worked as a mechnaical engineer in underwater remote controlled robot project• Desinged gripper arm that satisfies underwater working conditions using Fusion360• Advised software team about ROS and robotics software development• Assembled and repaired mechanical systems

• Developed and tested new computer vision features using C++ for a Real-Time AI assistant that enchants gaming experience of thousands of gamers.• Reduced code size by %15 by refactoring and improving C++ production codebase.

• Developed a GUI manual calibration tool to fine-tune LIDAR calibration parameters using; C++, ROS2, Qt

ITU Rover is a award winning team that manufactures autonomous space exploration vehicles for international competitions: URC and ERC• Prototyped custom navigation system for mobile robot that provides occupancy map generation, motion planning and PID control.• Implemented custom program to detect fiducial markers from RGB image• Experemented with visual SLAM ROS packages with stereo camera(ZED).• Prototyped Camera-LIDAR sensor fusion package to get depth images without depth/stereo camera• Labeled and prepared training images for our custom object detection models• Created various electronic systems and programmed microcontrollersUsed C, C++, Python, ROS, PCL, OpenCV