I’m a 3rd year CS student and I’m a bit stuck on where to go from here.

I feel like I can do algorithms and LeetCode reasonably well. I’ve learned a decent amount of DSA through college and I’ve built a few projects in different languages, so I don’t feel like I’m starting from scratch.

But I don’t feel like I’m a good programmer yet.
What I really want to get better at is being given a problem I’ve never seen before and actually being able to figure out how to solve it. Not just recognising a LeetCode pattern I’ve practised, but being able to approach something completely new and work it out.

I also feel like there’s just an insane amount of information to know. There are so many languages, libraries, frameworks, algorithms, tools, concepts, etc. that I don’t really know what I’m supposed to focus on anymore.

So for people who are experienced programmers, where do I go from here?

What did you actually do after you had learned the basic algorithms/data structures and could solve LeetCode problems that helped you become actually good at programming?

Should I focus on building bigger projects? Pick one language and get really good at it? Read other people’s code? Learn more about how computers actually work? Spend more time solving things without looking up the answer? Just start building things and learn whatever I need along the way?

And specifically, how do you get better at the problem-solving side of programming when you encounter something you’ve never done before?
I’m not looking for a LeetCode roadmap. I feel like I already know how to study algorithms. I’m more interested in what the next stage actually looks like and how to stop feeling overwhelmed by everything there is to learn.

A lot of what I'm doing when solving problems is actually not that different from your pattern recognition. It's just that I've seen a lot more code than you have, and in particular I've seen a lot of real world production code.

I'd say the best thing for you to do is try to make some more serious projects. While you're doing this, take the opportunity to study some real oprn source code related to your project. For anything you pick, you're almost certainly not the first person to do it, so go and see how others did it and learn from them.

Tbh tasks like LeetCode could be useful in recruitment, but not so much at work, nor they make you a good programmer. They teach you how to calculate stuff efficiently, which is useful, but not the issue in most cases. You need good coding practices. Learn about stuff like SOLID, domain driven development, hexagonal architecture. See how people comment and document the code, how to test it properly, this kind of stuff. What mostly separates a good programmer from a bad one is not what you can do on your own, but how can you cooperate in a team, which requires you to write readable, extensible and safe code.

Just keep finding things that interest you and work on them and when you do, don’t rush them. Slow down and understand every part of what you’re making to the point that you could explain it to someone else. Continue doing this, keep your passion going, and I guarantee everything else will fall in place.

This all comes from experience practical hands on experience . And one of the things especially is if the problem doesn’t matter to you in any way you just won’t care enough to resolve it . Working through exercises is great but I would honestly build something meaningful to yourself have the problems break things get angry research then resolve it. I promise you that kind of repetition is far more important than standardised exercises that are about as interesting as a someone wearing tweed and pointing at the board. (20+years coding)

The thing that changed it for me wasn't more algorithms. It was building something that had more than one user.

LeetCode teaches you to make code produce the right answer. Production teaches you to worry about code that produces the wrong answer without failing. Those are different skills, and the second one is most of the job.

Concretely: build one small app where two different people log in and must not see each other's data. Nothing fancy. You'll immediately hit questions LeetCode never asks. Who is allowed to do this. What happens when someone forgets a check. How would I even know it's working. What breaks when I add a second feature on top. That one constraint taught me more than another hundred problems would have.

The overwhelmed feeling doesn't really go away, for what it's worth. You just get better at knowing which parts you can ignore for now.

Build something, anything . Scratch an itch. You are very unlikely to find a problem someone else hasn’t fixed. There’s millions of developers on this planet and they’ve solved all the problems already.

Master the shell

I agree with this. Coding is a bit like art. The more you do, the more you'll be comfortable doing it, and the more challenges you'll come across. So go do some doodling, or go try to make something you like (games are usually fun).

This is actually a good real world project or problem for you. Try writing a server that uses a database, sql queries, secure login, user data, user server interactions that perform checks or actions, srikanth the proposed a very good project for you :)

For I started doing real projects instead of tutorials. Try building solutions

Thanks. The "checks and actions" part is the bit worth dwelling on it's easy to build the version where the check passes. The interesting exercise is deliberately trying to break your own rule and confirming the server says no.

I think (as much as the job market sucks) you should try to get an internship. I think getting an internship somewhere that has software devs (even if you aren’t actually explicitly a software dev intern) can provide the opportunity for you to talk to people who code in a real environment for a living, and (Maybe!) you can even ask your boss for a small project or to help THEM with a project... Ask your boss to set up a meeting where someone can tell you about the tools they use/have created for themselves. Even if your boss DOESN’T find you a project like that, try to find problems at your job that could be solved with programming

Or maybe put together a group of your CS friends and watch Wolf of Wall Street together and convince yourselves you can create the next big thing and spend 6 months learning how to manage a “real” small project with multiple devs.