![[Translate to Englisch:] Bachelorarbeit bei OCTUM](/fileadmin/_processed_/e/1/csm_bachelorarbeit_d14fdb37c9.jpg)
We are looking for a student to undertake a Bachelor’s thesis on the topic:
Data drift in AI systems used for image-based quality control
AI-based image processing offers great potential for pharmaceutical production. At the same time, in a highly regulated environment, AI systems must not only function reliably from the outset, but also remain stable over the long term.
This is because changes in ongoing production operations can occur gradually – for example, due to batch variations, slight product changes or ageing of the lighting. It is precisely such changes that can influence the performance of an AI system without this being immediately apparent.
And this is exactly where your Bachelor’s thesis comes in.
Among other things, you will investigate:
- typical causes of data drift in pharmaceutical production images
- suitable metrics and methods for detecting data drift
- methods such as image statistics, ROC-AUC or OOD scores
- the practical feasibility of AI-based image processing in production
- the implementation of a monitoring process using a sample project
The aim is to develop a practical approach that enables data drift to be detected at an early stage, assessed and communicated transparently.
In doing so, you will be contributing to a highly relevant topic: the responsible and consistently reliable use of AI in regulated production environments.
This topic is particularly well-suited to you if you are studying Computer Science, Data Science, Machine Vision, Automation Technology, Mechatronics or a similar degree programme – and are keen to combine theory with real-world industrial applications.
Interested in a bachelor’s thesis with practical relevance and a forward-looking topic?