Despoina’s journey into artificial intelligence began in Thessaloniki, Greece, where she grew up fascinated by how technology could change the world around us. While studying Electrical and Computer Engineering at Aristotle University of Thessaloniki, she encountered the world of AI through various projects, which instantly sparked her curiosity. Wanting to dive deeper, she moved to Amsterdam to pursue a master’s in Artificial Intelligence at the University of Amsterdam. Since then, she has been exploring the many ways AI can make a real difference, from research to practical applications. What she loves most about this field is how fast it moves; there’s always something new to learn, some fresh idea to explore, and that constant evolution keeps her inspired every day.
Despoina Touska is one of the two MSc AI students who were accepted into the AI Fellowship programme, which supports the education of young AI talent awarded based on academic and societal achievement, and sponsored by Qualcomm Technologies, the unit’s key industry partner. The AI fellows also received mentorship and guidance from Qualcomm’s top engineers. The Fellowship is made possible through the collaboration between Qualcomm Technologies, the ELLIS unit Amsterdam, the MSc AI Programme of the University of Amsterdam, and the Amsterdam University Fund (AUF). Let’s congratulate Despoina on her excellent achievement and graduation!
Tell us more about your MSc AI journey.
“I chose UvA for its strong research environment and international outlook, which felt like the right place to grow as a researcher. The AI master’s program stood out to me because of its research-oriented approach and the chance to learn from leading experts. Amsterdam itself, with its vibrant international community and reputation as a hub for innovation, made the decision even easier.”
During my master’s study, time was definitely limited, but I still tried to make space for other activities. I contributed to a project at a UvA lab focused on detecting toxic industrial emissions, where I primarily worked with Bayesian Variational Autoencoders. I also worked as a teaching assistant for the Interpretability and Explainability in AI course in the Master’s program, which was a valuable experience given the importance of the subject. Outside of academics, I enjoy swimming, and more recently, I have started running as well. I also really enjoy watching movies, reading, and spending time with friends and family.
What are your academic and career goals? Do you have any role models in this field?
Academically, I really want to dive deeper into computer vision and keep building my skills in areas like self- and unsupervised learning, vision–language, and multimodal models. I find it fascinating how these approaches bring together different ways of understanding the world — images, text, and beyond. I also hope to stay active in the research community, publishing and learning from others who are just as passionate about AI. Career-wise, I aim to grow into an AI Research Scientist role, though I also enjoy the hands-on side of developing and deploying machine learning systems.
I don’t think I can pick just one person, because several people have really inspired my work. One of them is Francesco Locatello from ISTA. His contributions to object-centric learning, especially Slot Attention, and his ideas around disentanglement and causality have had a big influence on how I think about inductive biases in vision models and even shaped parts of my thesis. I also really admire Max Welling. His work in probabilistic modeling and his ability to bridge academia and industry are something I find incredibly inspiring. It’s the kind of balance I’d love to achieve in my own career; staying grounded in research while working on things that have real-world impact.
Did you have a challenge you have overcome during your study?
One of my proudest achievements was my master’s thesis project, where I worked on applying AI to a complex real-world problem in semiconductor metrology. It was a project that required combining a deep literature review with careful engineering and evaluation to design a model that truly fit our requirements. The biggest challenge was that I was entering a domain that was completely new to me and not one that’s widely explored in the AI community. So I had to figure out how to translate the problem into the right set of techniques and justify each design choice. I ran extensive experiments and kept a close feedback loop with my supervisors to sanity-check key decisions, and in the end, we developed a solution that met our goals.
Can you explain more about the thesis project you conducted?
For my master’s thesis, I worked on a project in collaboration with a company, focusing on how AI can be applied to lithography metrology. In other words, using AI models to measure and analyze microscopic patterns on semiconductor wafers to ensure precision in manufacturing. After diving deep into the literature, I found that object-centric methods were particularly promising for our data, so I focused on adapting and experimenting with them in this context.
My main supervisor for the project was Pascal Cerfontaine, whose guidance was invaluable throughout the process. On the academic side, I also had insightful discussions with Dr. Martin Oswald and PhD candidate Tejaswi Kasarla from the University of Amsterdam, as well as with Dr. Lyubov Amitonova and PhD candidate Maximilian Lipp from ARCNL.
So, what is next for you?
“I view this degree as a foundation for turning ideas into real-world applications. I’d love to apply the AI fundamentals I have learned, especially in areas like computer vision, to develop new approaches and build solutions that are both innovative and practical. Beyond the technical side, this degree also connects me to a network of alumni and researchers. I see that as a huge resource; a community I can learn from, collaborate with, and stay inspired by as the field continues to evolve.”
After graduation, I would like to join a research or machine learning engineering team, working on computer vision or other areas. I really enjoy the full process, building models end-to-end, from gathering and understanding the data to training, evaluation, and deployment. I would also like to contribute to open-source projects or co-author papers with my team when there’s something new to share. It’s a great way to keep learning, give back to the community, and stay connected to the research side of AI.
It’s always tricky to predict five years, especially in a field that moves as fast as AI. But I envision myself staying involved in the community, continuing to learn, collaborating with other researchers and engineers, and working on projects that have a real impact. What matters most to me is staying curious and growing alongside the field, while contributing to work that pushes technology forward in meaningful ways.
How do you see the impact of the Qualcomm fellowship?
“I see the fellowship as a real force multiplier in promoting inclusion within AI. First of all, it opens doors; by providing financial support and mentoring, it helps talented people enter the field who might not have had the opportunity otherwise. That kind of access can make a huge difference. It also creates visible role models.”
When early-career researchers from underrepresented groups succeed, it sends a powerful message to others that they belong here, too. And beyond that, diversity strengthens the work itself; teams with different perspectives are better at spotting blind spots, building fairer datasets, and designing solutions that truly serve a wider range of people.
I think the program is really strong from a scientific point of view. It gives a solid foundation across all the core areas of AI and does a great job of incorporating recent advances, like foundation models, which I really appreciated.