AI will reshape software engineering, but it won’t eliminate it. The boring stuff, such as coding, routine QA, and documentation, is bounded and pattern-heavy, so AI will eat that first.

The real bottleneck has never been typing. It’s figuring out who the stakeholders are, what they need, and why. That’s messy, political, and brutally hard to automate. For most products, the critical work is defining the problem, not writing the solution.

That kind of work requires soft skills, requirements engineering, deep domain knowledge, and prompt engineering. It’s also much harder to outsource, because deep language and cultural awareness are critical.

If you want to future-proof your career, focus on being really good at understanding and defining the problem to be solved.

Spend less than you earn. Live well below your means. Do this for years and everything gets easier.

You can also just try to earn a shitload of money and not worry about it.

Personally I’ve found the live cheaply option to be much easier.

@AlanClifford got on my soap box before I had a chance to. But even if you ignore AI (don’t do that), the larger issue is that software development, especially generic enterprise development has been headed toward commoditization for years.

If all you have to sell is “I codez real gud” and “I pull well defined tickets off of a Jira board”, you’re screwed. No being able to reverse a b tree on the whiteboard while riding a unicycle on a tightrope won’t differentiate you.

"There is no better way to achieve job security than by making yourself an indispensable employee."

- Lou Bloom, Nightcrawler.

I'm already working on my own side hustles. future proofing yourself currently is bound to making sure that you have deep knowledge and can solve things that i can't.

The amount of generated code is getting more and more. But also, on the other hand the amount of production ready code is less than before. because people need to make sure it works which in my experience with vibe coding, it never works if you don't review the code apply strict constraints and a good modular folder structure.

Only startups and solo developers skip that part.

Formally build up your network.

It's not a question of "areas of focus" in the technical sense. I mean, be a worthwhile engineer, yes, of course. But "nobody hiring" is BS. Horror stories, yes, but still just about everybody is working. LeetCode is good interview skill, if you can do it under pressure and with a hostile audience (rather than fresh, in peace and in the morning.)

I find it shocking nowadays that people get laid off or fired and don't have dozens upon dozens of "buddies" they can remind themselves to. Not even after months of feeling like they might be next.

To build your network: be pleasant and useful to work with not just to your immediate colleagues. Instead be also pleasant, curious and available to everybody in sight at your company. Actively talk to them, ask for meetings and discussion and overviews of what happens around them. Among what you will hear about is all the soft skills that you may not already have. And all local vaguely related interest groups, and online ones, and trade shows, and dev conferences, etc, etc. Formal in that you might as well maintain a CRM-like database of this network. Use LinkedIn so that other people can find you, instead of just you finding them. Four years might be a little early, but cultivate headhunter relations also.

Look for advancement opportunities within your company. Don't necessarily get hired there but probably still talk to them. It's still more people who would love to know you.

If you feel "state school" may be a little insufficient... Four years is nearly enough that which school doesn't matter much anymore in your career. But you might be ready and IF you get laid off, you might do an MS at a "much better" school. This time choose one that will let you balloon your network with everybody else there that you might run into.