Working students at the Industrial Data Lab: refining skills in a real-life environment

Innovation, knowledge exchange, and development of associates and students – these are the goals of the Industrial Data Lab that has been established by Bosch and the University of Stuttgart. It is based on the Technology Partnership Lab Program of the University of Stuttgart and kicked off in October 2022. This cooperation offers students the opportunity to gain practical experience in different ways, be it as a working student, for a master or PhD thesis, or in research projects. A great opportunity for students to apply what they have learned during their studies, and exactly what the working students Ekaterina and Tobias are currently doing.

Ekaterina: gaining practical experience from Russia to France to Germany

Ekaterina has been a working student at Bosch Digital for almost six months now, gaining her first professional experience in Germany. She is currently completing her master’s degree at Faculty 5 of the University of Stuttgart with a focus on Natural Language and Speech Processing. This is a good opportunity to put the knowledge learnt in lectures into practice and apply Deep Learning (DL) methods to real-world scenarios. Ekaterina is mentored by a team of specialists at Bosch Digital. “The guidance from senior colleagues with a PhD in my field of interest is very valuable to me,” Ekaterina explains.

Her background includes pre-sales engineering for SAS Institute Russia/CIS (2018-2021), which involved customizing ML solutions for various industries in collaboration with colleagues around the world. During her one-year master’s program in France, Ekaterina conducted a small R&D project in computational semantics using PyTorch to extract entities from texts. Following this more scientific approach, she felt it was important to gain industry experience and applied to Bosch Digital.

“I’m currently involved in an NLP project focused on GDPR – identifying sensitive documents, anonymizing data, and doing multiclass document classification with few annotated samples. This is a great opportunity to explore few-shot learning techniques, such as SetFit, as well as gain hands-on experience in fine-tuning NER models and packaging them into user-friendly APIs while delivering value to an inspiring company like Bosch Digital,” Ekaterina summarizes.

Tobias: challenges call for creative solutions

Ekaterina’s colleague Tobias is currently studying Computer Science (M. Sc.) at the University of Stuttgart. “I already gained some experience in classical software development. In my bachelor thesis on ’A unified open- and closed-source software requirements dataset’, I got into contact with the topics of Neural Networks and NLP. These topics fascinated me,” Tobias recalls. “That’s why I decided to continue on this learning path and further deepen my knowledge in these topics.”

At Bosch Digital, Tobias is currently working on a project to automate the digitization of documents, focusing on training multiple classifiers for each document language. “In the first step, we train a binary classifier to allow or reject documents for storage. The second step consists in using another classifier to assign the approved documents to one of several possible categories,” he explains. “Several challenges arise here due to a wide range of different languages, very long documents, and a general variance in the data sets. Creative solutions need to be found, which is what excites me about this job. I’m glad to be part of this great team and look forward to future data projects here at Bosch Digital.”