Advising Statement

My goal as an advisor is to help PhD students and postdoctoral researchers become independent scientists with strong technical skills, good research judgment, and the ability to identify and pursue important problems.

This statement describes my approach to research and advising. Practical expectations about research ownership, feedback, meetings, professional conduct, projects, and career development are described separately in the Advisor–Advisee Expectations.

Research philosophy

Our group develops learning methods that combine modern machine learning with mathematical and scientific structure. Current themes include representation learning, optimization, implicit constraints, dynamical systems, and learning from simulations and synthetic data.

I value research that leads to a genuine conceptual or methodological advance. This often requires moving between mathematical reasoning, algorithm design, implementation, and experiments. Research is rarely linear: ideas fail, experiments contradict expectations, implementations contain bugs, and promising directions sometimes need to be reconsidered.

Curiosity, intellectual honesty, persistence, and good research judgment matter as much as technical ability.

How I advise

I provide scientific guidance, challenge assumptions, help identify promising directions, and give candid feedback. I enjoy discussing both high-level questions and technical details, especially when a project is at an important conceptual or methodological stage.

I often give pointers rather than complete solutions. Following a lead into the literature, deciding whether it applies, and coming back with a precise question are important parts of becoming an independent researcher.

My level of involvement depends on the researcher and the project. New students or researchers entering an unfamiliar area may benefit from closer interaction. As expertise develops, I expect my role to shift toward discussing, challenging, and refining ideas rather than directing each step.

I am also happy to discuss research positioning, collaborations, applications, and career plans when useful.

Independence and scientific development

The main goal of a PhD is to learn how to conduct independent research. Early in the PhD, I may be closely involved in defining questions, developing approaches, and interpreting results. Over time, I expect students to increasingly identify problems, propose ideas, make technical decisions, and shape their own research direction.

Independence does not mean working alone. Asking good questions, seeking feedback, and knowing when to ask for help are important research skills.

Postdoctoral researchers are expected to operate with greater independence from the outset, lead research directions, contribute new ideas, and develop a distinct scientific identity. They should also contribute intellectual leadership to the group when appropriate.

The European Commission’s ResearchComp framework provides a useful broader description of the competences researchers develop during this process.

Scope, collaboration, and advising fit

I can be most directly engaged in projects that overlap with my current research interests and expertise. Researchers are welcome to explore directions further away from my own work, but they should expect greater independence and may benefit from additional mentorship or collaboration.

Co-supervision can be especially valuable because it brings together different expertise, perspectives, and research styles. When several people are involved in a project, I encourage open scientific discussion and clear communication about the role of each supervisor or collaborator.

Research interests and working relationships can evolve over time. If the supervision or scientific fit becomes difficult, I prefer to discuss it early and consider, with the relevant people and institutions, what arrangement would best support the research.