PhD Position in ML Accelerator Programming for HPC and Scientific Computing

Join Vrije Universiteit Amsterdam (VU Amsterdam) as a PhD candidate in ML accelerator programming for HPC and scientific computing, and help develop methods and tools to make ML devices accessible and productive for Scientific and High-Performance Computing.

Your function

The growing computational demands of Machine Learning (ML) have led to the rise of spatial devices explicitly designed for ML workloads, such as the Cerebras Wafer Scale Engine, AMD Versal AI Engine, and Tenstorrent Blackhole.

Although these accelerators can offer performance and efficiency gains to application domains beyond ML (e.g., computational sciences, big data analytics, graph processing), their potential in these areas remains unexplored. This is largely due to the lack of comprehensive software ecosystems, making them difficult for experts to use and inaccessible to domain scientists.

In this PhD project, you will help bridge this gap by demonstrating best practices and by developing programming tools and methodologies that democratize access to ML accelerators and other spatial devices for the wider scientific community.

Your duties

As a PhD student, you will develop and lead original research. Your work will combine quantitative research, theoretical investigation, software design and development, and experimental validation. Specifically, you will:

• identify and characterize workloads and application domains that benefit from ML accelerators, understanding when and how they can accelerate diverse applications
• develop methodologies and open-source software tools for computing specialists to facilitate application development and reduce time-to-solution
• enable accessible programming for domain scientists, providing them with high-level tools and training resources to productively use ML accelerators without requiring deep hardware or low-level programming expertise

As part of this position, you will be encouraged and supported to refine and drive your own research within the project scope, publish and present your findings at top-tier systems and HPC venues, develop and maintain open-source software, and contribute to teaching and community outreach. The ultimate goal is for the successful candidate to produce a dissertation that leads to a PhD degree in 4 years.

Your profile

We offer a fully funded PhD position for 4 years. We are looking for a highly motivated candidate who meets the following qualifications:

• a Master’s degree in Computer Science or Computer Engineering (completed before the start date of the PhD)
• solid background and competencies in parallel programming, high-performance computing, performance engineering, and computing architecture
• excellent problem-solving and programming skills, with proficiency in C/C++
• prior knowledge of accelerator programming (e.g, GPU), and, particularly, of spatial computing architectures (e.g., AMD AI Engine, Tenstorrent) is considered a plus
• excellent oral and written communication skills in English
• strong organizational skills and the ability to work both independently and as part of a team
• proactive and enthusiastic about addressing cutting-edge computing systems research challenges

We realise that each individual brings a unique set of skills, expertise and mindset. Therefore we are happy to invite anyone who recognises themselves in the profile to apply, even if you do not meet all the requirements.

What do we offer?

• a salary of € 3.059,00 (PhD) and maximum € 3.881,00 (PhD) gross per month, on a full-time basis. This is based on UFO profile PhD candidate. The exact salary depends on your education and experience.
• an employment contract of initially 18 months. If there is sufficient perspective, this will be extended to a total of 4 years. Your dissertation at the end of the fourth year forms the end of your employment contract.
• 8% holiday allowance and 8.3% end-of-year bonus
• solid pension scheme (ABP)
• contribution to commuting expenses
• optional model for designing a personalized benefits package

About us

This position is funded under the “Machine Learning Accelerators for Science: Productive Spatial- Hardware Programming for the Scientific Community,” awarded by NWO (Dutch Research Council).

The candidate will be embedded in the Massivizing Computer Systems (MCS) group, which focuses on research in distributed computing systems and ecosystems, and currently spans over 40 diverse people, of whom 3 are staff. Dr. Tiziano De Matteis (the main supervisor) is an Assistant Professor in the group. His principal research interests are related to High-Performance Computing (HPC) with a particular focus on Systems and Programming models for Spatial and Reconfigurable computing for HPC, Parallel Programming, and Sustainability of Modern Computing Systems.

The MCS group (led by prof. Alexandru Iosup) has a central role in organizing the national and international computer systems communities and has played leading roles in organizing top-tier conferences (e.g., HPDC, ICPE, CCGRID) and journals (e.g., IEEE TPDS). The MCS group publishes in high-quality conferences, journals, and magazines, and has a long legacy in conducting grand experiments and in releasing FAIR data and software artifacts. The group has links with national and international organizations and plays a leading role in international academic-industrial partnerships, such as the SPEC Research Group and the LDBCouncil. Last, but not least, the MCS group has a sustained and leading presence in developing young talent through individual supervision and embedding in large, cross-disciplinary, international teams.

Faculty of Science

Working at the Faculty of Science means collaborating with students, lecturers and researchers who are focused on their field, yet have a broad view of the world. We are proud of a positive and inclusive workplace culture within the faculty, where we work together with great energy and a pragmatic attitude to tackle social challenges.

Vrije Universiteit Amsterdam

Vrije Universiteit Amsterdam stands for values-driven education and research. We are open-minded experts with the ability to think freely - a broader mind.