INTERVIEW
techniques, sub-30 fs temporal resolution and pulse-by-pulse spatial / phase characterization. Most recently, we began using fibers themselves as physical analog computing elements for classification tasks, a direction we are now extending to multimode fibers for higher dimensionality.
You also work on integrated waveguides. What is the objective there? We are now in the process of transferring part of what we learned in fibers to integrated platforms such as thin-film lithium niobate or tantalate. These systems offer strong nonlinear and electro-optic effects over very short distances, enabling compact devices with lower power requirements. Our goal is to determine whether key nonlinear fiber phenomena can be replicated, or re-engineered, on-chip for future integrated photonic applications.
Which emerging topics do you find particularly promising? Two lines of inquiry stand out. The first is optical computing using nonlinear media, whether in fibers or integrated waveguides, to act as analog processors. This is attractive, although I remain cautious: it is beautiful physics, but it may not replace GPU-based computation. The second is nonlinear self-organization in multimode systems, enabling passive shaping of spatial and spectral profiles simply by tuning input power and conditions. Overall, the unifying theme is structured light across temporal, spectral and spatial dimensions using nonlinear control.
What about modelling and numerical complexity? In single-mode systems, well-established models accurately describe nonlinear propagation. In multimode systems, however, simulation becomes dramatically more complex: one moves from one-dimensional fields to three-dimensional spatiotemporal data cubes. Simulating long propagation distances with high resolution becomes computationally prohibitive, especially when scanning parameters. This is one reason why we increasingly combine experiments and machine learning instead of relying solely on brute-force numerical modelling.
What are your main collaborations? Locally, I collaborate with several groups at our university. Internationally, my longest-running collaboration is with Prof. John Dudley in the Louis and Marie Pasteur University in France. I also collaborate with groups in Sapienza University in Italy, the Institute for Photonics Technologies in Germany and a group in Chengdu University in China.
Are you using photonics to support machine learning or machine learning to support your investigations in photonics? Both directions are now active. Initially, machine learning served photonics by helping to analyse and control nonlinear dynamics. More recently, we have also used photonic systems themselves as hardware for analog computation. Stabilization of experiments is another area where machine learning can be valuable using feedback algorithms to compensate for drifts and maintain alignment during long data acquisitions.
How has the photonics ecosystem developed in Finland, particularly with the national flagship? Until mid-2000, Finland was not a leading EU country in photonics research despite solid infrastructures in Helsinki, Tampere and Eastern Finland. Three key developments changed the landscape: significant university investments in fabrication technologies; the transformation of the Finnish Optical Society into Photonics Finland integrating industrial stakeholders; and the Academy of Finland’ s call for eight-year flagship programmes. By unifying the main national academic groups and leveraging industrial momentum, we secured a national photonics flagship in 2019, now involving nearly 500 researchers, with an application for an eight-year extension underway.
What has been the impact of the flagship? The flagship has increased research quality, strengthened funding, accelerated innovation( more patents and start-ups), expanded training programmes, multiplied scientific and industrial events, and significantly improved international diversity. Finland now hosts more than 300 photonics companies for a population of 5.5 million( which is one of the highest densities in Europe) creating strong demand for highly trained graduates.
What about doctoral education reform? PhD studies in Finland were considered too long and insufficiently connected to industry. National pilot programmes were launched two years ago to shorten doctoral timelines and increase industry placement. We secured one of these programmes for photonics, and now coordinate a national doctoral network with about 70 PhD students aligned with the flagship and industrial needs.
You also coordinate a national infrastructure network. What does this involve? Our fabrication and characterization platforms were labelled national infrastructures of strategic importance, meaning they are open to researchers and industry across the country. We provide access, expertise, and support for design and prototyping. This complements the flagship and the doctoral network, forming an integrated ecosystem of research, training and innovation.
How did EOSAM come to be organised in Tampere? EOSAM is the annual conference of the European Optical Society, whose headquarters are in Finland. I already knew the EOS coordinator, Elina Koistinen, and during discussions I suggested hosting the conference in Finland in Tampere. EOS lacked the manpower to organise it alone, but with the flagship administrative resources, budgetary support and network, we could take on the local organisation. The proposal was submitted and accepted, and EOSAM 2026 will take place in Tampere on August 2026.
What guides your approach to research? What matters most to me is the articulation between fundamental research, genuinely useful applications, and strong investment in training and mentoring. If this contributes to strengthening photonics in a small country like Finland, then it is a very positive outcome.
18 www. photoniques. com I Photoniques 137