We use GitLab. It has CI/CD integrated. It also has container registry[0] and package registry[1], so you can have your Docker images and packages for npm, Go, PyPI, etc.
You can also build images in your pipeline with Kaniko[2], and there are some useful things like pages, either private or public, for your different repositories. Here's an example docs https://big-mama-tech.gitlab.io/bmc/
It does secrets managements, deployment tokens, etc. You have everything in one place, and you can also integrate it with your Kubernetes cluster, say with Google Cloud Platform, for example.
All the best, this sucks. On a slightly related note, we are building our machine learning platform and one thing we're working on right now is to get everything out of it: meaning the notebook servers you spawn will be on your own cluster that you give authorization for, the data will be on the cloud provider of your choice or your cluster, the experiment notebooks will be there, the long-running jobs will be on your cluster, the models, the model images, the deployed moels, pretty much everything, precisely to avoid that kind of scenarios where a service shuts down and then users are left to scramble to exfiltrate their work and output that span many years from that service, looking for ways to do it and "Download as a ZIP file" crap. If the service ever shuts down, everything should be on the user's existing infrastructure. We have the plumbing set, and we're working to provide porcelain for this to users. One guideline for us is to be able to die without users being impacted [which is also a guideline for me, being able to die without the team being impacted]. So we're giving a lot of thought to this because it sucks when it happens.
- [0]: https://docs.gitlab.com/ee/user/packages/container_registry/
- [1]: https://docs.gitlab.com/ee/user/packages/
- [2]: https://docs.gitlab.com/ee/ci/docker/using_kaniko.html