Spend that time figuring out how to change it to unit tests in your IDE, or making the deployment lighter weight.

Work on the meta-problem that makes your problem easier. After all, you've got time if the baseline is low-productivity. And the lessons learned are portable.

Not everything can be unit tested. Figuring out what ide is doing and writing a plugin for what you actually want to do.. How to maintain plugin across ide updates...it's the start of rabbit hole and may not be worth it sometimes..

Break out of the IDE then -- run your unit tests from the command line. As an added bonus, it will be much easier to troubleshoot the headless CI builds too.

You're overestimating how much control I've got over the company-wide infrastructure.

Unit tests don't really fix the problem either. I'll have integration tests that run automatically during builds either way because I want to make sure other people are not breaking my stuff and there's no real difference in time between setting it up for 1 test or 1000.

The only solution I've found so far is try to bypass as much of the company-specific infrastructure as I can during development, but that requires learning a lot of docker/k8s/helm on top of knowing the company-specific stuff well enough to know what I can ignore on top of rake libraries that ~~abstracts~~ obfuscates the infrastructure.

Learning the k8s stuff I don't mind that much because as you say it's portable, but the company specific stuff... I doubt it's even portable across the company.

5 minutes is great, try 3 weeks (due to various "objective issues" aka fuckups across the organization).

5 minutes is actually on the low end. I put it more as a minimum that really disrupts my flow. Below 5 minutes is kinda tolerable... 5 minutes to 5 hours... not so much.

You have my deep sympathy. That sounds crazy!