In the summer of 2022 I worked at Scansonic, a company within the Berlin Industrial Group that develops laser processing and sensor systems for automated joining in automotive manufacturing.

ML QA detection from 2:00 onwards

ML QA detection from 2:00 onwards

Ground Truth Annotation Tool

While there, I worked on their AI inference team who use deep learning to detect welding defects for the purpose of quality assurance. I helped improve their ground truth annotation software, leading to a reduction in labelling time of ~15%, by incorporating the following:

• Integrated an internal library to increase the speed of data loading times by a factor of 20
• Implemented an automated test suite for their inference software
• Integrated automated coverage analysis for the inference software with +80% coverage