Résumé

Computer scientist from Kempten in the Allgäu. By day a software quality consultant, on the side I build my own software and run a 3D printing business.

Work

CQSE GmbH↗

Software quality consultant. I analyse clients' software systems, make technical debt visible and help teams make quality a shared goal in their daily work. Articles and talks on teamscale.com.

3D Bruckner↗

My small side business in Kempten (Allgäu): I 3D print one-off parts and small runs, and handle everything myself, from modelling to shipping.

Education

M.Sc. Computer Science: Games Engineering

Technical University of Munich. The programme combines computer science with computer graphics, simulation and game development. I focused on artificial intelligence and robotics.

B.Sc. Computer Science: Games Engineering

Technical University of Munich. The undergraduate foundation in computer science, complemented by graphics, simulation and game development.

Theses

Written at the TUM Chair of Robotics, Artificial Intelligence and Real-Time Systems. Learning-based methods can replace classical control in robotics, but they usually learn only a single task. I used hypernetworks to train reinforcement learning agents on several manipulation tasks in sequence, based on state representations that are themselves learned continually from camera images. In simulation the agents do not forget earlier tasks, and the image-based representation matches or beats hand-crafted state vectors.

Bachelor's thesis↗

Written at the TUM Chair of Software & Systems Engineering. Using suffix-tree clone detection on source code comments, I found many typical patterns that code generators write into their files. They became a repository of generator patterns that automatically classifies code as generated or manually maintained. I tested it on a reference data set and a large collection of open-source projects to determine the share of generated code.

Guided Research↗

“Automated Camera Stabilization and Calibration for Intelligent Transportation Systems”, in the Providentia project at the A9. Camera-based traffic infrastructure has to locate detected objects accurately in the three-dimensional world, and keep doing so in continuous operation. The work calibrates cameras automatically from targets in high-definition maps and compensates for fast vibrations and slow orientation drift with a stabilization mechanism. It was evaluated on a real-world test site.

Publications & Talks

IEEE IV 2024↗

“An Online Self-Correcting Calibration Architecture for Multi-Camera Traffic Localization Infrastructure”. L. Strand, M. Bruckner, V. Lakshminarasimhan, A. Knoll. IEEE Intelligent Vehicles Symposium 2024, pp. 1666 to 1671. The paper grew out of the Guided Research and describes the calibration architecture that corrects itself during operation.

Workshop 2025↗

“Qualität für alle und alle für Qualität” (German), with Dr. Tobias Röhm, 28 May 2025. In the workshop we showed how teams build a shared picture of software quality and shared goals through regular quality retrospectives. With a live demo of how to prepare one, and customer examples.