THESIS PROJECT: Hardware acceleration of machine learning

Background
Connected vehicles in Volvo Group continuously stream a wealth of diagnostic data from on-vehicle sensors. Utilizing such data in combination with machine-learning models to predict component failures ahead of time would enable us to increase part availability in our dealer network, reducing customer downtime.


Thesis questions and expected outcome
The wealth of diagnostic data requires increased computing capacity to train, optimize and deploy ML models. Hardware accelerators such as graphics processing units (GPU) and field-programmable gate-arrays (FPGA) are increasingly being used to bypass such computational constraints.
The purpose of this MSc thesis is to benchmark and optimize the performance of different hardware accelerators in the training of ML models related to predictive maintenance of vehicles. Examples of possible optimizations include exploring hardware features such as SIMD instructions in processors and lower precision operations in GPUs.
The results will allow the efficient parameter optimization of ML models for the prediction of component failure and facilitate the deployment and practical use of the models within Volvo group for the purpose of predictive maintenance.


Student profile and application
MSc student in Computer Science, Computer Engineering, or other related discipline. You have an interest in hardware acceleration and machine learning
Application deadline: TBD, we will continuously review the applications so don’t wait with submitting.


Contact information
For further information, please contact:
Recruiting manager:
Thomas Nordenskjöld at Advanced Analytics Europe
E-mail: Thomas.nordenskjold@volvo.com
Phone: +46739025355


Volvo Group Mentor
Evangelos Siminos, Data Scientist at Advance Analytics Europe


E-mail: Evangelos.siminos@volvo.com
Phone: +46 765536464


Kindly note that due to GDPR, we will not accept applications via mail. Please use our career site.