• Development of Phoenix, a Software Service for training and evaluating AI models (for image based tasks) in the cloud

• Implemented a solution for robotic grasping of unknown objects leveraging a CNN in PyTorch applied on Depth Camera data• Developed an open-source framework for distributed Deep Reinforcement Learning on robots based on ROS, Python and gRPC https://github.com/jr-robotics/robo-gym • Developed Gitlab CI to automate the test of the robot simulations in Docker containers and the release of pip packages• Managed connections with national and international Machine Learning communities• Implemented a Gazebo simulation of an agricultural machine• Developed a mission assignment system for an intralogistics system

• Deep Reinforcement Learning applied to Robotics• Mobile Robots control• Robot Simulation

Worked on the Master Thesis with the title: "Task allocation and planning for heterogeneous multi-agent intralogistics teams"

Student job during my studies. Over the years, I have worked in almost all required positions, showing high flexibility. To name just a few of the jobs held: Cashier, order taking, customer service, kitchen help and pizza making. In the process, I have learned how to handle stressful situations in a calm and focused manner and how to deal with customers in any situation.

Plug And Navigate ROBOTS for smart factories (PAN-ROBOTS) is a Seventh Framework Programme project. Its objective is to develop, demonstrate and validate a generic automation system for factory logistics in modern factories based on advanced Automatic Guided Vehicles (AGV). PAN-ROBOTS consortium proposes a new generation of flexible, cost effective, safe and green AGVs in combination with advanced infrastructure systems. Those advanced AGVs will be able to transport material and products in modern factories based on autonomous on-board path planning and navigation to enable high flexibility. The perception system to guide the AGVs through the factory will be based on a novel cooperative approach. Advanced on-board sensors will be combined with infrastructure sensors to enhance the cost effectiveness and increase safety. The fleet management will be intuitive and easy to use by workers. In addition the installation time and costs will be dramatically reduced by semi-automated plant exploration supported by a localization approach utilizing already existing landmarks and an advanced pallet handling system which detects and picks the pallets autonomously. Finally, the developed generic system will be exemplarily validated in the production process of a bottling company and against the needs provided by the already established PAN-ROBOTS user group.