I’m a computer vision & robotics software engineer who builds perception systems that let robots see and act in the real world. My work centres on 6‑DOF pose estimation, deep‑learning and classical vision pipelines, and the distributed, scalable software that runs them in production.
I hold an M.Sc. in Computational Sciences in Engineering from Technical University of Braunschweig, specialising in robotics, computer vision, and artificial intelligence. Core courses included Robot Control & Optimisation, Pattern Recognition, and a hands‑on Robotics Laboratory. Alongside my studies I researched fault‑tolerant edge robotics at Intel Labs, co‑authoring a preprint on stateful failure recovery for distributed ROS 2 systems.
Across research groups and industry, I’ve built perception software, behaviour‑tree testing tools, and large‑scale 3D visualisation platforms. My goal is to engineer reliable systems that let robots understand their surroundings and act with confidence.
Away from the keyboard you’ll find me travelling, reading or discovering new music.
Experience
Aug 2025 – Present Computer Vision & Robotics Software EngineerUnchained Robotics GmbH
- Full-Time
Own end-to-end development of deep-learning and classical perception pipelines for 6-DOF pose estimation of mixed, occluded items, using industrial 3D cameras and a DETR-based instance-segmentation model selected for its occlusion handling. Built the classical OpenCV pipeline from scratch on Intel RealSense RGB-D data, including eye-in-hand and eye-to-hand calibration, reaching pose accuracy within 5 mm and 1° in continuous 24/7 production. Generate physics-based synthetic data in Isaac Sim for sim-to-real transfer, and enhanced the core Python framework acting as the software PLC for robotic-cell orchestration.
- Python
- PyTorch
- OpenCV
- DETR
- Instance Segmentation
- 6-DOF Pose Estimation
- Intel RealSense
- Isaac Sim
- Hand-Eye Calibration
- gRPC
- FastAPI
Jan 2024 – Jan 2025 Robotics Software EngineerIntel Labs
- Internship
- Master-Thesis
Contributed to Scenario Execution, an open-source library that couples OpenSCENARIO 2 with Behavior Trees for reproducible robotics experiments. Added automated tests and GitHub Actions CI, and built Kubernetes-based container workflows that offloaded computation to remote nodes, cutting onboard resource usage by 80%. For my master thesis, designed a fault-tolerance framework enabling a mobile manipulation system to recover statefully from mid-task failures in around 1.5 seconds.
- ROS2
- OpenSCENARIO
- Behavior Trees
- Kubernetes
- Container Orchestration
- GitHub Actions
- MoveIt2
- Nav2
- Python
- C++
Mar 2023 - Dec 2023 Robotics Research AssistantInstitute of Mobile Machines & Commercial Vehicles, TU Braunschweig
- Research
Built a ROS 2 framework for sensor-data exchange across a multi-machine system (drone, ground robot, edge, cloud), using image-chunking to reliably transfer large captures over a 5G network, and configured multi-machine ROS 2 discovery. Integrated real-time GPS streams for precise positioning during field trials.
- ROS2
- Python
- C++
- 5G
- GPS
Feb 2022 - Feb 2023 Full-Stack DeveloperLower Saxony State Office for Monument Preservation
- Working Student
Engineered a React, Node.js platform that manages and visualises giga-scale LiDAR point clouds of historic buildings. Integrated Potree for in-browser 3D rendering and tuned Postgres with indexing & partitioning to accelerate spatial queries.
- React
- Node.js
- TypeScript
- Tailwind CSS
- PostgreSQL
- Potree
Projects
Fault-Tolerant Edge Robotics: Stateful Failure-Mitigation Framework
Reactive, Behavior-Tree-driven framework that detects application and communication failures in distributed ROS2 manipulation and navigation stacks, then restores operation on a Kubernetes edge cluster.
- ROS 2
- Kubernetes
- Behavior Trees
- OpenSCENARIO 2
- Gazebo
- MoveIt 2
- Nav2
- Prometheus
- Docker
Realistic Failure Scenarios for Underwater IMU Navigation
Time-based fault-injection framework that simulates realistic IMU and magnetometer failures inside an underwater ROS/Gazebo environment to stress-test AUV navigation algorithms.
- ROS1
- ROS2
- ROS1 Bridge
- Gazebo
- C++
- Docker
- IMU (9-DOF)
- AUV simulation
HistoDot — Web Platform for Architectural Heritage Data
Web platform that centralises architectural-heritage datasets, offering search, metadata editing and in-browser 3D point-cloud visualisation. Designed for researchers and conservators to easily explore, annotate, and share large scans.
- Svelte
- TypeScript
- Node.js
- Tailwind CSS
- PostgreSQL
- Potree (point-cloud)
- Docker
