Graduation Project: Auto Data Evaluation

The worldwide operating company ROSEN inspects several assets using self-developed techniques and devices. As part of ROSEN R&D, the sensor data gathered by inspecting tank bottoms with high resolution Magnetic Flux Leakage (MFL) technology are processed to recognize possible anomalies by the help of machine learning. For this field of expertise, they’re looking for a graduation student (Computer Science/Data Science) to further improve existing algorithms. 

Lees hieronder de hele vacature

Vacature details

Geplaatst op
Aantal uur

Functie omschrijving

By the help of two available labeled and unlabeled datasets, it is aimed to test several methodologies to the one hand improve the accuracy of the current classifier and on the other hand generalize the model toward data variations caused by different factors. 


More specifically the following research topics should be covered in this project: 

  • Which benchmark and metric can optimally determine the baseline? 
  • What optimization/normalization techniques can improve the current algorithm quality? 
  • Is there any benefit of using metric learning for higher quality class distribution modeling? 
  • How to use feature extractions such as PCA or Auto-encoders to reduce the complexity of system? 
  • Is the data quality improvement by further samplings and evaluations, effective?


To become part of the ROSEN family, you should bring with you: 

  • Having background of machine learning, preferably, classification and clustering
  • Good python programming skill (C++ programming skill is a plus) 
  • Pro-active and team-oriented attitude interested in working with large data


We offer various career development opportunities of an international, innovative and sustainability-oriented company. In an open corporate culture with rapid decision-making, you can implement your ideas successfully. Moreover, you can expect enthusiastic support by experienced engineers and scientist, working in a dedicated team with a ‘can-do’ mentality. 

We attach great importance to a balance between work and family life. Moreover, flexible working hours and different working time models are standard for ROSEN.

Bedrijfsprofiel ROSEN Group

Previously, potential anomalies were first detected directly from the raw measurement data and then manually crosschecked by the evaluators in order to select only the corrosion spots; here a lot of efforts were needed. Recently, machine learning techniques such as classification, based on existing ground truth databases, relatively successfully have been developed by pre-classifying the potential anomalies. Nevertheless, a further algorithm quality improvement is wished. 


By clicking the application button, you will be navigated to the website of the company. In case the link is not working anymore, the vacancy has expired and you will no longer be able to apply. We try to keep our job database as up-to-date as possible, we would very much appreciate it if you could let us know in case a link is not working. 

Would you like to be kept up-to-date about other interesting jobs in Twente, then sign up here and create your own job account

Meer informatie

Zutphenstraat 15
0541 - 671 000