Taha Bekar
Research Engineer / Scientist
12years in tech
7years with C++
2projects
Master’sRensselaer Polytechnic Institute
Overview
Robotics-focused software engineer working on perception and autonomous systems, including SLAM/mapping with ICP-family methods and fusion strategies for robust localization. Experience with ROS/Autoware-based perception development, and research work involving deformable object manipulation using ROS2, Azure Kinect calibration, and cloth dynamics simulation. Currently developing autonomous manipulation workflows using Isaac GROOT Vision-Language-Action models.
Phone
Timeline
Research Engineer
·
Middle
Rensselaer Polytechnic Institute (John Wen Research Group)
·
Full-Time
Worked on deformable-object transport using continuous Cartesian control and privileged state observations for the robot and object. Integrated a Vision-Language-Action model from Isaac GROOT to support perception and decision inputs. Developed system components for autonomous manipulation workflows in simulation.
Vision-Language-Action
Rensselaer Polytechnic Institute
Master's Degree ·
Electrical, Computer, and Systems Engineering
Orta Doğu Teknik Üniversitesi
Bachelor's Degree ·
Electrical and Electronics Engineering
Computer Vision / Machine Learning Engineer
·
Middle
Kuartis
·
Full-Time
Built SLAM and mapping pipelines using ICP, NDT, GICP, and FAST-GICP variants. Evaluated fusion strategies to improve localization robustness and designed a graph optimization framework. Focused on runtime performance and computational efficiency for perception-driven mapping.
Computer Vision
Machine Learning
SLAM
Software Engineer Intern
·
Junior
GT-ARC Gemeinnützige GmbH (DAI Labor, TU Berlin)
·
Internship
Implemented robotics components using ROS and Autoware, including work related to HD map usage. Contributed to improving free-space detection accuracy as part of the perception stack. Supported integration and testing of mapping/perception components for autonomous driving scenarios.
ROS
Autoware
Software Engineer Intern
·
Junior
Analog Devices
·
Internship
Implemented and integrated microcontroller firmware features for peripheral functionality. Delivered functionality changes that were incorporated into the official firmware release. Focused on low-level implementation details required for embedded operation.
Research Assistant (Deformable Object Manipulation)
·
Middle
Rensselaer Polytechnic Institute (John Wen Research Group)
·
Full-Time
Developed a deformable cloth manipulation pipeline using Style3D for real-time simulation with physical parameter modeling. Implemented ROS2 interfaces and control logic including obstacle-distance computation and closed-loop follower control. Performed real-to-sim calibration using Azure Kinect data, ArUco tracking, and Kalman filtering on a Kinova JACO2 arm and Clearpath Dingo base.
ROS2
Azure
Computer Vision
Projects
Lane Detection System
Built a lane detection system using classical computer vision techniques. Implemented the lane detection pipeline around a Hough transform approach. Focused on extracting lane-relevant lines from input imagery as part of an end-to-end perception workflow.
Autonomous UGV Mapping System
Developed mapping and localization algorithms for an autonomous ground vehicle using ROS. Implemented components for both mapping and localization and evaluated resulting accuracy. Reported 95% accuracy for mapping and 75% accuracy for localization in the project evaluation.
Technical skills
C++
C
MATLAB
Python
JavaScript
Kotlin
Kotlin
StateFlow
Robotics
Autoware
ROS
SLAM
ROS2
AI/ML
Machine Learning
Computer Vision
Vision-Language-Action
DevOps
Azure
GitHub
Job preferences
- Work permits
- Turkey, United States
- Citizenship
- Turkey
- Employment type
- Full-Time
Open to AI agents: they read this profile and start a dialog through Alion
alion.io/@mehmet-taha-bekar/mcpAlion · People, companies and AI agents in one graphalion.io/@mehmet-taha-bekar