Ask HN: What problems do you face in your machine learning work?

Hi, Assuming you do it professionally, how do you go about your machine learning projects? What is your workflow? What tools do you use and why? What tools don't you use and why not? How is your team organized if you have a team? How do you collaborate and hand-off each piece of the puzzle? How do you handle data, develop your models, track experiments, deploy and manage your models, etc. I'd love to hear your thoughts. Disclosure: we're building a platform[0] for our own needs for now, but I'm interested in how other people do it. Right now, we have real time collaborative notebooks, no-setup huge images with the most popular libraries, long-running notebook scheduling with output that survives closed tabs and network disruptions, automatic experiment tracking [we automatically detect parameters, models, and metrics and save them], AppBook publishing (one click parametrized notebooks), model deployment, and packaging models in Docker containers you can push to a registry. Thank you, - [0]: https://iko.ai