PhonicUK

Not much actually. I use a local accounting firm that take care of the businesses tax affairs, payroll, and general reporting to the appropriate authorities.

Licencing is fully automated and I never have to touch it after I built my licence system. An order comes in via Paypal, my servers are notified, a licence is generated and sent to the user along with a formal PDF invoice. No problem. The same system can also be queried for VAT receipt reports each quarter.

The only real pain that I wish someone else took care of is the categorization of expenses. There's a spreadsheet in a particular format that goes to the accountant for each financial year, and it's a slightly slow process getting all of the outgoing expenses into it and putting them all into the right category. It's probably actually something that could be automated, but its only something that has to be updated every few months (taking about 2/3 hours) so I can't really justify the time cost building software to do it.

What I really bemoan is a proper universal 'invoice' file format that's always issued as standard. There's still a manual process to copy data from invoices into something more machine handleable, be it a spreadsheet or dedicated bookkeeping software.

Build something that solves a problem you have and for which there either isn't an existing solution, or the existing solution(s) aren't suitable (cost, quality, etc). Once you've got the 'guts' working to solve your own problem, make it pretty, make it friendly, then sell it.

Even if you don't make any money off it, at least you solve your own problem in the process.

No. On a personal level, I don't consider the US as somewhere with a suitable social safety net if things went south. I'd never live anywhere without universal healthcare for example.

On a professional level, it'd be a highly expensive move that wouldn't be justified by any of the benefits (which I'm really struggling to see). So to a large degree I wouldn't see the point in making such a move.

This is tricky actually because I started such a long time ago that I don't know where I'd suggest to start if you were getting into it for the first time today. I've been learning to code (and you never stop learning) for over 21 years now and a lot has changed.

What I'd say is that the language(s) you chose to learn aren't as important as a solid understanding of the theory involved. Although I will say that learning a language that uses static typing instead of dynamic typing will prevent you learning some really bad habits.

I'd also suggest the project based approach. Decide on something you want to build, and learn what you need as you go along. Make lots of mistakes, throw away all your work several times and be prepared to invest a lot of time. Unfortunately programming as a skill has a really long ramp up time between day 0 and basic competence.

Mine had the advantage in that I had software I'd already made and was well known in its community before the company existed, and as the associated community grew so did awareness of my software. So I got to skip the whole reputation building problem.

Never stop learning would be my tip.