JupyterHub allows you setup research clusters on GCloud, AWS and Azure. You can set CPU / GPU resource utilization limits, disk usage, memory, network. Even limit scaling to your budget. Once your experiment is up and running. It's simply another service running in a container. Have used it for a small distributed team. But can be scaled to corporate R&D teams with 1000s.
Core environment is still the Jupyter Notebook. So should remain familiar to most data scientists.
Zero to JupyterHub with Kubernetes