Ask HN: Have you successfully done a career do-over, and how did you do it?

It sounds like you were away from coding for a really long time (10 years!), then you got back into it, picking an area you didn't know much about, expected it to be easy, maybe didn't put in the time outside of work...this approach wouldn't work for anyone, no matter their age.

Here's a manager-y question for you: how could you have done things differently when you went back to being an IC, to have had a better transition? Maybe doing more prep work before you left the manager track? Maybe going back to C++ instead, or an area that was closer to what you used to work on? Maybe doing open source work in that new java/python target area, to get experience, build a portfolio, and be sure you liked it, before you bet your livelihood and reputation on it?

You need 10 years of tensorflow experience for this.

Okay that is helpful. In my experience, this situation -> people that, for one reason or another, appear to wander around with no clear career progression or direction, so it looks like they are "junior" in experience despite being "senior" in total years worked. has always been that the person wandering didn't know what they wanted, and if it had been years, didn't have tools to figure out what they wanted.

I am reasonably certain there are other explanations/causes but the three cases I can recall easily were all that the person got into a career in computer programming because someone one else told them they should. And they stayed employed but they didn't have any idea about what they wanted to do so they changed jobs for all sorts of reasons, a girlfriend, a pay raise, a manager that kept bugging them at the old job. All of the change reasons were external to their career path.

One really wanted to be a musician. They could talk on and on about different styles and influences. They lit up and were on fire. One was lost, they had always had their life planned out for them by others and they followed that plan until it ran out (just after graduate college and get a job). And they never previously had been required to create their own plan. I didn't get to talk with the third person I can remember because they moved on to a new position before I had that chance.

If that resonates with you then my advice is to work on figuring out what you really care about. And I recognize that isn't an easy task, it was simple for me I was fascinated by computers and the systems you could build with them, so much so I build stuff just for fun. But a good way to search for the things that you care about are to explore different things (very hard to do as a single breadwinner in a family). As I told my musician friend, even if you're not an artist with a signed label contract, you have a understanding of what they are going through and you can write code. Can you find ways to help other musicians that way? Or music lovers? Or screenwriters trying to match music to scenes? or hospitality businesses wanting to influence the mood with music? You can get quite meta and still exploit the experience you've developed in an area you care about.

Without an amnesia treatment there is only so much you can reset :-) Mostly folks learn basics about work (its not always fun, how you work can be as important as what you get done, your boss may be an idiot but it doesn't change the fact that they are your boss, etc). So instead of 'resetting' it's more like vectoring. Now that you've been in the career for a while, if you know what you'd like to be then imagine you have reached that point and and then try to imagine plausible steps that you would have had to take to get there.

Often, it's the situation that architects your decisions. If you can delay debts and dependencies, you can afford to do more. Live within your means. Downsize your needs. Practice frugality.

>After 10 years of that I wanted to move back into pure coding. With great experience should be easy right? Wrong. The trouble is that the "just coding" experience gets very quickly outdated. Your knowledge of Win32 API or WPF isn't really relevant on a Python job. Interestingly, the skills that don't get outdated mostly lay outside the technical part: your ability to manage people and resolve conflicts, your ability to present ideas and convince others, your ability to spot business niches and shape products under limited resources.

In the old times one could argue that being a good engineer (i.e. finding simple solutions to tough problems) makes a difference, but I'm not sure how much this holds anymore for software engineers, since the complexity isn't in about fitting your business processes into the SQL/backend/frontend bounds, but rather in defining those processes and making proper assumptions about the market.