2017: Training outstanding doctoral students at IMPRS-IS – Interview with Sofiane Ouaari

This month, we’re looking back to 2017 when Cyber Valley’s graduate school, the International Max Planck Research School for Intelligent Systems (IMPRS-IS), was founded. IMPRS-IS trains outstanding doctoral students from around the world and supports them in their PhD research so they can use their AI expertise to build a better future in Europe.

Sofiane Ouaari is a student at IMPRS-IS. He joined the graduate school in 2022 where he works with Nico Pfeifer in medical informatics. His research focuses on healthcare data, ensuring that machine learning pipelines protect this sensitive information.

In the following interview, Sofiane explains how IMPRS-IS has supported his development as both a researcher and an entrepreneur. 

 

What initially drew you to IMPRS‑IS, and what convinced you that it was the right environment for your PhD?

During the final year of my Data Science Master's degree, I set my sights on pursuing a PhD. I wanted to deepen my knowledge in machine learning and push myself beyond my comfort zone. Research felt like the ideal environment for that kind of growth: challenging, exciting, and full of opportunities to develop.

I first heard about International Max Planck Research Schools (IMPRS) through a contact, and once I looked into the ‘Intelligent Systems’ branch, it felt like a natural fit for my background in data science and machine learning. The application process itself was a good first impression, everything was clearly explained, both in writing and through videos, which made the whole experience smooth.

What ultimately convinced me was the scale and richness of the network. The connection between Tübingen and Stuttgart, the diverse set of faculties and associated faculties, and the international community of PhD students all pointed to one thing: a genuinely vibrant research environment. It had all the right ingredients to excel and to properly and efficiently kick-start a research journey.

 

From your perspective, what is unique about IMPRS‑IS’s culture and research community? What distinguishes it from other doctoral training programs?

What makes IMPRS-IS stand out for me is how much it offers beyond the research itself. It's almost like an n-in-1 program, where n represents the diverse number of workshops, training sessions, social events, and networking opportunities all layered into the experience alongside the actual PhD work.

One thing I'm particularly glad to have been part of is the organizing team for Think & Link, a monthly event that alternates between the MPI-IS campuses in Tübingen and Stuttgart. The concept is simple but highly effective: the ‘Think’ part gives PhD students a platform to share and discuss their research, while the ‘Link’ part is a fun social activity designed to break the ice and build real connections. And it works! IMPRS-IS isn't just a doctoral program, it genuinely feels like a community and, in many ways, a family.

 

What real-world impact do you envision for your research? Are there areas where you imagine it could make a difference in the future?

Data is the fuel that powers any AI system. If you think of AI as an engine, data is what keeps it running. But while data enables AI models to learn and generate real value, it also carries risk: sensitive information about individuals can leak through the very models trained on it.

That's where privacy-preserving machine learning comes in. It addresses the security layer of that engine, and it's becoming increasingly critical in a world shaped by regulations like GDPR and the EU AI Act in Europe, or CCPA and HIPAA in the US.

Any company building intelligent systems today needs to think carefully about how they handle personally identifiable information, but also how robust their models are against attacks like membership inference, which can reveal whether a specific person was part of a training dataset, or model inversion attacks, where an adversary can reconstruct user data simply by querying a model.

My work focuses on healthcare data, which raises the stakes even further. The EU AI Act explicitly classifies AI applications in healthcare as ‘High Risk’ and rightly so.

This makes privacy-preserving solutions not just a nice-to-have but an essential part of any responsible AI pipeline. Ultimately, this is what drives me: building machine learning systems that are not only powerful, but safe and trustworthy.

 

Looking ahead, what directions are you considering after your PhD, and in what ways do you feel IMPRS‑IS is preparing you for that next step?

It's genuinely hard to give a definitive answer right now, and I think that's partly because we're living through such a fast-moving moment in AI. Technologies like agentic AI are reshaping the landscape almost in real time, and the opportunities available in a year or two may look very different from today.

That said, whether I head into industry or pursue a postdoc, one thing I know is that I want to work on applied, real-world problems. I find it deeply motivating to see the concrete impact of my work and how it benefits others. IMPRS-IS has helped me stay grounded in that mindset while also giving me the tools, network, and broader perspective to navigate wherever that path leads. One concrete example of this is the S4 workshop series offered by IMPRS-IS. A particularly valuable session was ‘Successfully Apply for Industry Positions in Germany’, which gave me a much clearer picture of how to navigate the German job market and tailor my applications effectively.

Beyond workshops, there were also opportunities to visit industry directly. We had the chance to tour the BOSCH research and development center in Renningen, where we even presented our work to their team. That kind of direct interaction with industry is something you rarely get in a traditional doctoral program, and it really bridges the gap between academic research and the professional world.

 

IMPRS-IS is the graduate school of Cyber Valley, a network of leading research institutions, industry partners, and start-ups working in AI and robotics. How has being part of this network influenced your development, whether in your research, skills, or career direction?

Being part of Cyber Valley opened a dimension of my PhD I didn't fully expect and one I'm grateful for. I participated in the Cyber Valley AI Incubator as part of its fourth cohort, an intensive eight-week program that took us through the full journey of turning a research idea into a real start-up: defining a vision, analyzing markets, developing business models, building prototypes, and ultimately pitching to potential investors. Our team ended up winning the public prize of that edition, which was a proud moment.

More broadly, this experience reinforced something I now believe strongly: research and business aren't opposites, they're complementary. IMPRS-IS and Cyber Valley together push you to step outside the research bubble and think like an entrepreneur, to ask not just ‘is this interesting?’ but ‘does this create value?’ That shift in mindset has been one of the most valuable things I've taken from this journey so far.

 

To celebrate Cyber Valley's ten-year anniversary, we'll be sharing a variety of content across our website and social media channels reflecting on milestones from the last decade. Stay tuned for the next installments!