PhD Position in Storage Support for Machine Learning

VU University of Amsterdam Department of Computer Science


Are you interested in building the next-generation of efficient and fast computing systems using cutting-edge storage and networking technologies? Do you like building real distributed systems and evaluate them at scale? Do you want to work with ambitious colleagues working at the intersection of storage technologies, machine learning, and high-performance distributed systems in the exciting city of Amsterdam? Then we are eager to get to know you. Please apply for this PhD position at Vrije Universiteit Amsterdam (VU).


FTE: 0.8 - 1

Job description

The rise of Non-Volatile Memory (NVM) storage technologies like Flash and Optane, is one of the most fundamental changes in the storage systems since the advent of Hard-disk drives (HDDs) back in the 1960s. Today NVM storage can deliver 10s GBps bandwidths, millions I/O requests/sec, ultra low data access latencies (~a few microseconds). The rise of these storage devices has fundamentally reshaped how we build modern storage and distributed applications. 

One of the data-heavy and popular applications is Machine learning (ML), which is shaping many aspects of our daily lives such as healthcare, daily commute, finance, and IT infrastructure. ML applications rely on big datasets, which are processed iteratively to build a knowledge model. Most of the modern ML data processing and training happens from DRAM (both, CPU or accelerator attached) memory. As the size of training data and model increases, so are their requirements for the amount of DRAM needed. However, DRAM technology itself is facing pressure from multiple fronts - it is not scalable, is energy inefficient, and is costly.

Taking inspiration from the rise of NVM storage with their performance and open interfaces like OpenChannel and Zoned SSD devices, the key research theme that this work explores is how can we leverage modern high performance storage devices to reduce DRAM pressure from DRAM-heavy distributed ML workloads. In this work you are to explore questions like what are the performance requirements of ML workloads, can we do storage stack specialization to support specific workloads and access patterns, how do you deliver predictable performance with shared network and storage devices, how to do automatically synthesize placement and scheduling policies, what are the performance, cost, and energy advantages and disadvantages of using NVM storage for a distributed workloads, etc. In the process of answering these questions you will design and implement a specialized distributed storage system. 

Your duties

  • Analyze the literature on storage systems, distributed systems and systems support for efficient ML workloads using fast NVM storage and high-speed network APIs
  • Design and implementation of a distributed, shared storage service for ML workloads to support large-model ML training efficiently and its integration with a distributed ML framework
  • Distributed systems building, implementation, and conducting reproducible research following state-of-the-art reproducibility guidelines (additional help available)
  • Teaching responsibilities include supporting pertinent courses (e.g., storage systems and advanced computer programming courses, see here: and and supervising BSc and MSc thesis works in the scope of these ideas


  • A Master's degree in Computer Science, Computational Science or related field
  • Excellent programming skills (e.g., C/C++, Python, Rust, Go)
  • Strong collaboration and communication skills. The principal language of the group is English (fluent written and spoken English skills are required)
  • [Optional Plus] Background in low-level systems programming and projects related to operating systems, kernel programming, file systems, networks, storage, GPU programming, etc.
  • [Optional Plus] Experience with machine learning frameworks like PyTorch and TensorFlow

What are we offering?

A challenging position in a socially involved organization. The salary will be in accordance with university regulations for academic personnel and amounts €2,395 (PhD) per month during the first year and increases to €3,061 (PhD) per month during the fourth year, based on a full-time employment. The job profile: is based on the university job ranking system and is vacant for at least 0.8 FTE.

We especially encourage women applicants. In the case of equal suitability of candidates, preference will be given to a woman candidate. 

The appointment will initially be for 1 year. After a satisfactory evaluation of the initial appointment, the contract will be extended for a duration of 3 years. The starting date is flexible, but preferred dates are Sep or Oct 1st, 2021. 

Additionally, Vrije Universiteit Amsterdam offers excellent fringe benefits and various schemes and regulations to promote a good work/life balance, such as:

  • a maximum of 41 days of annual leave based on full-time employment
  • 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 Vrije Universiteit Amsterdam

The ambition of Vrije Universiteit Amsterdam is clear: to contribute to a better world through outstanding education and ground-breaking research. We strive to be a university where personal development and commitment to society play a leading role. A university where people from different disciplines and backgrounds collaborate to achieve innovations and to generate new knowledge. Our teaching and research encompass the entire spectrum of academic endeavor – from the humanities, the social sciences and the natural sciences through to the life sciences and the medical sciences.

Vrije Universiteit Amsterdam is home to more than 26,000 students. We employ over 4,600 individuals. The VU campus is easily accessible and located in the heart of Amsterdam’s Zuidas district, a truly inspiring environment for teaching and research.


We are an inclusive university community. Diversity is one of our most important values. We believe that engaging in international activities and welcoming students and staff from a wide variety of backgrounds enhances the quality of our education and research. We are always looking for people who can enrich our world with their own unique perspectives and experiences.

The Faculty of Science

The Faculty of Science inspires researchers and students to find sustainable solutions for complex societal issues. From forest fires to big data, from obesity to medicines and from molecules to the moon: our teaching and research programmes cover the full spectrum of the natural sciences. We share knowledge and experience with leading research institutes and industries, both here in the Netherlands and abroad.

Working at the Faculty of Science means working with students, PhD candidates and researchers, all with a clear focus on their field and a broad view of the world. We employ more than 1,250 staff members, and we are home to around 6,000 students.

About the research group

We are a part of the Computer Systems (CompSys) group,, at the Computer Science Department at VU Amsterdam which has a long history of doing world-class systems research. We have an on-going collaboration with many industrial and academic partners. 


Are you interested in this position? Please attach the following: Please apply via the application button and upload your curriculum vitae with the other needed documents (see below) until June 15, 2021. 

Needed documents: 

  • A motivation letter (max 2 pages)
  • A CV, detailing your technical expertise, past projects, and any relevant work experience
  • A list of courses and grades at the MSc level
  • A copy or draft of your MSc thesis
  • Names and emails of two references

Applications received by e-mail will not be processed.

Vacancy questions

If you have any questions regarding this vacancy, you may contact:

Name: Dr. Animesh Trivedi
Position: Assistant Professor
Telephone: +31 20 59 82219

No agencies

In your application, please refer to