Meet McGill University

McGill University is one of Canada’s most prestigious institutions and a globally recognised research university, with one of the highest proportions of international students of any Canadian university. 

This month, we spoke with Oana Balmau, Assistant Professor at the School of Computer Science at McGill University, who collaborates with INESC TEC on systems optimisations for Artificial Intelligence/Machine Learning (AI/ML), storage tiering, and, more recently, energy efficiency optimisations for AI. 

What is the history of collaboration between McGill and INESC TEC?   

The seeds of this collaboration were planted when Ricardo and I were both PhD students. He reached out about one of my projects on scheduling for key-value stores, and we have stayed in touch ever since. That initial connection led to a series of visits in both directions, which gradually grew into a long-term research collaboration. Over the years, we have worked together on three main research directions: systems optimisations for AI/ML, storage tiering, and, more recently, energy efficiency optimisations for AI. The common thread that has sustained and strengthened this collaboration is our PhD students. We are currently co-advising two PhD students, and their work continues to drive new ideas and joint projects. 

What have been the key enablers of effective co-supervision across institutions? 

One key enabler is regular communication. In an international collaboration, it is not possible to simply “bump into each other” in the hallway, so maintaining frequent contact is essential. In our case, this has been made easier by our weekly meetings with our co-supervised students. 

The INESC TEC International Visiting Researcher Programme has also been instrumental in strengthening the collaboration. One of our co-advised PhD students, Rahma Nouaji, secured a position through the program, and visits have also been arranged for me and one of my former postdocs. Similarly, our co-advised PhD student based at the University of Minho and researcher at INESC TEC, Rúben Adão, has been able to spend an extended research visit at McGill. 

Finally, collaborating on shared research grants has helped create long-term momentum. For example, João Tiago Paulo and Ricardo Macedo are collaborators on McGill’s NSERC CREATE grant, providing a common framework for joint research activities and student supervision. 

How has the McGill and INESC TEC collaboration across our joint research on storage, ML training, and energy efficiency enabled us to jointly submit papers to top venues? What do you think are the key ingredients for continued success? 

In addition to close communication that keeps the momentum going, I think the biggest factor behind our success has been complementary expertise. Our teams bring different strengths to the table, allowing us to tackle research problems from multiple perspectives and build stronger solutions together. 

Access to complementary hardware resources has also been a major advantage. By combining the infrastructure available at McGill and INESC TEC, we have been able to evaluate our ideas at a scale that would have been difficult for either institution alone. 

More broadly, I think sustained success comes from having aligned research interests, investing in long-term relationships rather than one-off projects, and creating opportunities for students to work closely across institutions. 

In your view, how important is it to prioritise research on storage optimisation, training optimisation, and energy efficiency over the next few years? 

I think this is an increasingly important research direction. As AI and ML deployments continue to grow, there is much greater emphasis on scalability, efficient resource management, and the cost of operating these systems at scale. Optimising model training and improving energy efficiency are becoming just as important as achieving higher model accuracy. 

At the same time, storage remains a fundamental challenge. Modern AI models are increasingly data-hungry, placing significant pressure on storage systems and data pipelines. 

How do you rate your experience with INESC TEC? What do you value most in this collaboration? 

This has been a wonderful experience overall, and I look forward to continuing our collaboration in the years ahead. Working together has always felt natural and productive, with a shared commitment to tackling ambitious research problems and supporting our students. 

What can we expect from McGill over the upcoming years? 

Stay tuned for exciting research on energy efficiency for AI/ML pipelines.