Cloud-based data analytics relies on a set of high-level user-friendly software and tools. The underlying middleware often rely on advanced software frameworks based on containers such as Kubernetes and allow interactive and real-time modes. While at the other end, High-Performance Computing relies on efficient low-level software implementation using C(++)/MPI and are used in a batch-mode relying on workload schedulers such as SLURM. Bridging the 2 works is an open challenge that is enabled by next generation of supercomputers which are highly heterogenous allowing a wide variety of modules such as cloud or even quantum modules to interoperate in a seamless manner.