At Quansight we’ve been hacking away at a new tool called conda-store and conda-docker to help manage software environments on Kubernetes without a docker dependency. We also have an alpha JupyterLab extension so that end users can easily build environments. In a separate project, we are building some ansible scripts that make it very simple to deploy a small HPC system with JupyerHub etc built-in on bare metal or ontop of something like vmware and we are modifying conda-store to work with that (will be open sourced soon). Just realized it would be a fairly small step to modify conda-store to also output singularity containers. Essentially it would automate going from a yaml file declaration of an environment to the singularity container. Is there much of a need for this?