Robotics & AI Engineer · Stuttgart, Germany
M.Sc. Computer Science (Autonomous Systems) · University of Stuttgart

I am a Master's student in Computer Science (Autonomous Systems) at the University of Stuttgart, focused on building robotic systems that are perception-driven, reliable, and deployable in real environments. My work sits at the intersection of robot perception, motion planning, and AI — with hands-on experience taking projects from simulation through to hardware execution.

Over the past year I have been working extensively with the Franka Research 3 (FR3) collaborative robot, developing a vision-guided manipulation pipeline that integrates multi-camera point-cloud fusion, Contact-GraspNet pose estimation, MuJoCo simulation, and differential IK execution. I have iterated this pipeline using a structured benchmark harness, improving grasp success rates from 38% to 90%+.

My technical interests include:

• Robot Perception — 3D LiDAR odometry, visual SLAM, point-cloud processing
• Manipulation & Grasping — grasp pose estimation, motion planning, sim-to-real transfer
• Locomotion — quadruped control, legged robot simulation (MuJoCo, ROS 2)
• AI/ML for Robotics — vision-language models, multimodal retrieval, deep learning for perception

I aim to contribute to serious robotics engineering efforts, whether in research or industry.

A production-quality pick-and-place pipeline for the Franka Panda robot that combines:

• Contact-GraspNet for 6-DoF grasp pose estimation from point clouds
• Multi-camera point-cloud fusion for robust scene representation
• MuJoCo simulation for policy validation before hardware deployment
• Differential IK execution for smooth, collision-aware motion
• A benchmark harness used to iteratively tune the pipeline from 38% to 90%+ grasp success

Python · MuJoCo · Contact-GraspNet · ROS 2 · PyTorch

An Eigen-only core odometry library with a ROS 2 Humble wrapper, supporting both offline and live RViz evaluation. Validated on a Boston Dynamics SPOT robot and handheld LiDAR hardware. Includes Docker support for reproducible deployment.

C++ · Eigen · ROS 2 Humble · Docker · LiDAR

Simulation and control stack for a quadruped robot performing manipulation tasks in MuJoCo. Explores locomotion stability combined with arm-based interaction tasks.

Python · MuJoCo · Pinocchio

End-to-end autonomous navigation stack for indoor environments, covering mapping, localization, and task-level autonomy.

Python · ROS · Navigation Stack

A fault-tolerant, decentralized chat application built from distributed systems fundamentals — message reliability, causal ordering, leader election, and dynamic scalability.

Python · Distributed Systems

A differential drive robot modeled in Fusion 360, exported to URDF, and simulated in Gazebo with full ROS control integration — including teleoperation and sensor integration.

ROS · Gazebo · URDF · CMake

A LiDAR sensor simulated from scratch in Python — covering ray casting, distance measurement, and 2D map rendering — built to understand the underlying physics before using off-the-shelf libraries.

Python · Sensor Simulation

A MATLAB simulation of a multi-joint robotic arm using Denavit-Hartenberg (DH) parameters, covering forward kinematics, workspace visualization, and trajectory planning.

MATLAB · Robotics · Kinematics

Open to research collaborations, full-time robotics engineering roles, and interesting open-source projects.
Connect on LinkedIn · View Resume