How to Run Python in Production

I thought that Ashish’s article had some interesting insights on running Python in production. Below are some of my thoughts. Keep in mind that I don’t have massively large Python projects in production. The tools that I created are internal to my company and only my dashboard runs all the time.

Project Quality

I generally agree that uv is a good place to start. I use uv for Python version management, project management and one-off scripts. I mostly use ruff for my linting and have been able to reduce the number of linters. I also make use of pyproject.toml as much as possible to config my linters as well.

Again since my projects are internal, I haven’t really checked for secrets though I will research Git Leaks and Nosey Parker a try.

Project Maintainability

More or less agree and I didn’t know about License Check. I will definitely need to look at this for my personal projects. I would also add that it is important to have good documentation. Documentation is not just a ReadMe, it includes documented objects in the program and a Wiki if the program is in anyway more substantial than a script.

Deployment

I don’t use Docker at work. Initially it was interference with my VMs, then it was confusion with the pricing in the past when used commercially though it does appear that I could use it now at work. Note at work I am forced to use Windows and most of those problems are not issues using Linux-based or Mac systems. Again my projects are not massive and using uv and virtual environments are good enough for me.

Conclusion

Good article and I did learn a couple of nice things. As always reading how others do things, take what works for you. Not everyone is working at a relatively small company like myself creating only internal tools and nor everyone works at a huge company like Google that need to create internal/external programs at scale.