UNESRALE

A programme by UNESCO and the Brazilian Center for Research in Physics (CBPF) gives researchers from resource-limited countries remote access to AI computing infrastructure, mentorship, and participation in joint scientific tasks — that is, access to infrastructure and expertise rather than funding. The laboratory is preparing three individual research applications within a shared method-first frame: we develop machine-learning methods, and specific domains serve as a testbed for validating them. Three directions: unsupervised anomaly detection in energy-equipment signals (energy); data-efficient fine-tuning of materials foundation models with synthetic-data generation (materials science); and physics-informed anomaly detection in flows based on known physical laws (fluid dynamics). For the lab the project combines scientific work with capacity building — training young specialists, postgraduate students, and students, and gaining experience with high-performance computing and open-science practices; results are planned to be open (code and datasets). The project is currently at the application-preparation stage.