We have two openings for self-funded thesis-based MSc candidates to begin by September 2024. Applications will be reviewed in early November 2023. This position is in data-centric systems to support vector-oriented computation
Supporting high-performance computation on vector-based workloads
The recent explosion of Machine Learning and Artificial Intelligence has made large language models and vector embeddings ubiquitous. These are numeric representations of objects in high-dimensional (e.g., 3d) space. The embedding space is learned such that objects that are similar to each other are also represented by vectors that nearby with respect to angle. Similarly, in Computer Graphics, salient points within a scene are often represented and compared with high-dimensional vector descriptors (such as DAISY) that enable finding common points across images during a 3d reconstruction pipeline.
What is common here is the need for highly efficient mechanisms for storing and processing high-dimensional vectors and performing complex analytics on them such as low-latency k-nearest neighbour search. Probably, the design of clever data structures (e.g., database indexes) that natively support these analytics could make many machine learning and computer graphics applications much cheaper, more energy-efficient, and more effective.
In this position, you will explore vector databases from a data structure design perspective. Through the process of designing novel, competitive data structures optimised natively for high-dimensional vectors, you will help design systems that enable more high-performance machine learning model deployment.
We are looking for creative, analytical, smart team members who are committed to self-improvement and want to push scientific boundaries. You should probably have (or expect to have before starting) an undergraduate degree in an Engineering discipline (such as, but not necessarily, Computer Science) from a very strong university or at least two years of industry experience writing fast and/or multithreaded code.
If you apply to this position, please try to position your strengths and/or interests with respect to the following:
- Can you explain clearly how data layout influences computational performance?
- Can you develop software that allocates and deallocates memory resources effectively?
- Can you design algorithms to solve a problem and demonstrate that they are correct?
Our gender-balanced, multinational research team consists of one Assistant Professor, two PhD researchers, four master’s students, and one undergraduate scholar, collectively from five countries. We are growth- and excellence-oriented: we have weekly skill development sessions and master’s students should strive to publish their research at top conferences. I meet at least weekly in a one-to-one context and am heavily involved in the research itself. We also communicate asynchronously with collaboration tools outside of those meetings. The team meets at least once per month for social activities. Researchers typically have access to a shared office inside the research lab if they plan to work in-person on campus. I hire in small cohorts and the intent is to have three researchers start and learn together in Fall 2024.
The university is situated near beautiful coastal old growth rainforest on an island that is slightly larger than Belgium. The temperature is moderate with only a week or two of snow in the city each year. It is common to enjoy skiing in the mountains in the winter, hiking and water sports, whale watching, swimming in the many large lakes in the city, the outsized variety of restaurants and cafes of a mid-sized capitol city, and expansive beaches. Canada is a liberal country in which you are strongly encouraged to become who you want to be.
Formally, you must follow the process outlined on the department website prior to receiving an offer.
Informally, you must also:
- contact me directly by email, positioning yourself with respect to the three questions above that characterise a successful candidate for this position. At a minimum, you should provide a CV, unofficial transcript, and three to five paragraphs explaining how you fit the candidate profile described above
- This step should be done prior to midnight UTC-7, Sunday, 5 November 2023.
The timeline is as follows:
- in the second full week of November (week number 45), I will contact shortlisted candidates who have the strongest evidence that they fit the profile above and meet the formal requirements of the university
- in the third full week of November (week number 46), we will conduct technical interviews with the shortlisted candidates
- in the final week of November (week number 47), we will conduct second round interviews and extend an offer
What is a Thesis-Based MSc?
It is important to note that at the University of Victoria, we have two options for master’s degrees: a thesis option and a project option. Our lab is only hiring in the thesis option this year.
The thesis option is focused on a novel two-year scientific research investigation that culminates in a thesis that you must defend in front of a committee to graduate. The coursework is auxiliary and consists of only five courses that are complementary to your concurrent research. The thesis option is necessary if you intend to pursue further graduate studies, such as a PhD, and highly advised if you wish to pursue a career in research and development (R&D).
The project option requires eight graduate-level courses and a lighter one- or two-semester research project. Generally, it involves less interaction with the supervisor as less research guidance and mentorship is required until the final two semesters.
Students in both options are encouraged to spend one or two four-month semesters in paid research internships or work placements to improve post-graduation employment prospects, develop additional skills, and to earn income to help cover study expenses. They are also encouraged to take part-time positions as teaching assistants to provide additional experience and income.