Thesis Work - Software abstraction layer for neural network accelerators i Göteborg

Thesis Worker at Volvo Cars

Welcome to explore the world of Volvo Cars by writing your thesis with us! As a thesis worker in our organization you are supported by a supervisor who follows you during your project. All thesis projects are arranged in business critical areas and therefore you will be able to contribute to our company purpose – providing freedom to move in a safe, sustainable and personal way – from day one! 

Designing and evaluating a software abstraction layer for heterogenous neural network accelerators

Background 
At Autonomous Mobility Platform we are building a winning team focusing on a high-performance computer with a distributed software, designed to enable innovation that takes us to fully autonomous driving cars. In other words, we create the platform foundation and architecture allowing consumption of all sensor data, decision making and actuation, on which our partners within and outside of Volvo Cars are building their mobility apps on top. 

Deep Neural networks are at the heart of enabling autonomous drive systems such as self-driving cars. They are used within different areas such as computer vision or Lidar based perception. However, DNN inference is a compute intensive task and to be able to timely execute DNN calculations, hardware acceleration is used. One such acceleration technology which has traditionally been used, are General purpose GPUs. More recently there has been a trend towards task-specific hardware accelerators, optimized for neural-network inference execution. 

To be able to cater for different types of hardware accelerators from different vendors, which also might change over time, it is of high value to abstract the hardware to enable the core neural networks algorithms to be deployed across this heterogenous hardware with minimal effort. There have so far been several industrial consortium driven initiatives for defining an abstraction layer and a few prominent vendor proprietary standards.

Scope 
The purpose of the thesis is to make an initial survey of prior work within the field and then design and implement a proposal for an abstraction layer of Deep Neural Network accelerators from different Hardware vendors. To limit the scope a suitable subset suitable for a sample DNN will be the focus. The proposed abstraction layer should then be evaluated and compared with using the hardware accelerator natively. The evaluation could be done of a number of criteria such as:
•    Inference Performance
•    Adaptability, how easy is it make an adaption for a new hardware
•    Expressiveness, how easy is it to express a wide range of neural networks.

Do you fit the profile 
We are looking for a pair of students in a Master program in Computer Science or Computer Engineering. Meriting fields of interests are machine learning with a focus on neural networks, computer architecture and systems programming. Prior knowledge of C and/or C++ is required.

We are flexible to explore possibilities to conduct the thesis remotely, however candidates for this position should preferably have the possibility to be present in our Gothenburg office on somewhat regular cadence.

Duration 
•    Start in January 2022
•    30 points ECTS (academic credits) if in agreement with your Thesis Advisor in University 
•    Number of students: 2

How to learn more and apply 
Selection will be ongoing during the application period, so do not hesitate to send in your application. Attach your Cv and personal letter stating your interests within the given area and your thoughts and credentials. Apply as soon as possible but no later than October 31st. Please note that applications via email will not be accepted.

For questions about the thesis work please contact Alexander Robertsson at alexander.robertsson@volvocars.com.

Who are we?

Everything we do starts with people. Our purpose is to provide freedom to move, in a personal, sustainable and safe way. We are committed to simplifying our customers’ lives by offering better technology solutions that improve their impact on the world and bringing the most advanced mobility innovations to protect them, their loved ones and the people around them. 

Volvo Cars’ continued success is the result of a collaborative, diverse, and inclusive working environment. The people of Volvo Cars are committed to making a difference in our world. Today, we are one of the most well-known and respected car brands, with over 40,000 employees across the globe. We believe in bringing out the best in each other and harnessing the true power of people. At Volvo Cars your career is designed around your talents and aspirations so you can reach your full potential. Join us on a journey of a lifetime as we create safety, autonomous driving and electrification technologies of tomorrow.