Benedikt Seidel

• Implementing an end-to-end Deep Learning Approach combining Graph Neural Networks and Convolutional-NNs (PyTorch, PyTorch-Geometric, ROS)

Scene Understanding in Lidar Perception using Machine-Learning: •Implementing state of the art methods in Pytorch for Self-Supervised-Pretraining of Lidar-Point-Clouds •Implementing Object detection metrics (AP-score, mAP etc.) with pytorch_lightning

• Implementing Oauth2-proxy with ingress files and connecting it with Azure App registrations to use a Kotlin app. • Working with Azure, Docker Kubernetes, Helm charts for apps built with Kotlin. • Working in an agile backend team that specifies in building NLP pipelines.

• bensbier.com, a craftbeer company creating creative beers

The focus of this Master is Machine-Learning in its various forms, highlights include: • Review Paper (Seminar Work): Graph Rewiring to improve Message Passing Neural Networks • Practical Project: Pose estimation between consecutive LiDAR scans using Pointnet++ vs Gradient based ICP + photometric optimization in PyTorch • Practical Project: Programming a fully autonomous quadcopter simulated environment (Unity) using ROS and C++ based on the DARPA (SubT) Challenge

A Bachelor in Biomedical Engineering made up of Electrical engineering, Data science, Software Engineering, Physical- and Biophysical fundamentals. Thesis Topic: Structured Adversarial Rewiring for Robust and Sparse Neural Networks - Institute of Theoretical Computer Science TUGraz (Supervisor: Lis. MSc Ph.D Ozan Özdenizci; Professor: Univ.-Prof. Dipl.-Ing. Dr.techn. Robert Legenstein)