The next generation of giant scientific infrastructures, such as the High Luminosity LHC and SKA, will produce exabytes of raw data each year. The optimal exploitation of their potential is hampered by the traditional paradigm that separates calibration, reduction, and scientific analysis of data into slow, offline processes.

Furthermore, the separation between computing and storage makes the application of deep learning techniques to the data stream difficult and inefficient, relegating them to post-hoc analysis. This gap between observation and discovery significantly limits the scientific yield of instruments designed as unique infrastructures shared by an international community.

In this presentation, I will introduce ODISSEE, a project uniting the efforts of three pan-European ESFRI infrastructures (the HL-LHC, SKAO, and SLICES-RI) with the aim of providing scalable and efficient hardware and software platforms to tackle the corresponding digital challenges in a highly competitive international environment. Developed as a joint R&D program with industrial partners, and thanks to access to SLICES-RI's experimental infrastructure, ODISSEE is designed to provide unique solutions along the digital continuum taking into account the 3 dimensions of sustainability.