Ask HN: How do you improve as a self-taught programmer?

Aside from that just keep doing what you're doing and revisit your old code every so often, you'll be surprised by how bad it is and you'll always find ways to improve it.

As for learning. IME reading and experimenting is the only way to do it. Yes, structured learning will go a long way, but even than, doing is the best teacher. You'll do something and make it work than 3-4 months later look back at it and be like, OMG that is so horrible I cannot believe I did it that way. The reality is that is true for all of us when we are learning, and frankly after 20+ years I can still look at stuff I wrote say last year and be like damn, I could do better today. New grads coming out of school require a few years of experience to become proficient and honestly it isn't their education that is bad, it is just that doing is the best teacher many times. I spend a lot of time showing young devs designs and things only experience can teach you. I have also seen new grads come out and they obviously have been spending their time experimenting cause they freaking rock it from day one, so experience & experimenting is what matters.

As for selecting tools, systems etc. My advice is stick with what is proven most at this point since that will typically have the most support on places like Stack Overflow and in general forums. This will also mean you can get design ideas and assistance from books and online resources easier and faster than if you try to adopt bleeding edge tech. For databases, stay with SQL for now, postgres or mysql are both fine choices, I'd favor postgres but honestly that is personal preference. If you are in a Microsoft stack and have those tools, stick with MSSQL and Visual Studio, great tools and tons of resources. For coding, stick with common frameworks and languages, Python is totally solid. Understand too that a tool that is awesome for backend work, or computational work etc may not be the best tool for UI/Web work. For example, Python is awesome as a backend tool but I wouldn't recommend putting together Web UI's with it. You can as many teams do, but the level of effort and lack of common resources makes it harder than say using Javascript/node.js or even Ruby.

For algorithm, curve fitting, regression type stuff, don't forget to check out Khan academy, and also check out youtube there are some awesome people posting great content on there (just pay attention to recommendations here cause some youtube content is crap too).

For me, carefully reading a language manual while thinking of a specific project was hugely successful. You must thoroughly understand the language. From your uncertainty it sounds like you should review your chosen language and master it.

Clean Code: A Handbook of Agile Software Craftsmanship [0] is a great book on writing and reading code.

Similarly, Clean Architecture: A Craftsman's Guide to Software Structure and Design [1] is, no surprise, a book on organizing and architecting software.

Designing Data-Intensive Applications [2] may be overkill for your situation, but it's a good read to get an idea about how large scale applications function.

The Architecture of Open Source Applications [3] is a fantastic free resource that walks through how many applications are built. As another comment mentioned, reading code and understanding how other programs are built are great ways to build your "how to do things" repertoire.

Finally, I'd also recommend taking some classes. I started as a self-taught developer, but I've since taken classes both in-person and online that have been a tremendous help. There are many available for free online, and if in-person classes work better for you (motivation, support, resources, etc), definitely go that route. They're a fantastic way to grow.

[0]: https://www.amazon.com/Clean-Code-Handbook-Software-Craftsma...

[1]: https://www.amazon.com/Clean-Architecture-Craftsmans-Softwar...

[2]: https://www.amazon.com/Designing-Data-Intensive-Applications...

[3]: http://aosabook.org/en/index.html

Style your Python as well as your SQL

Cleanup field/variable naming

(i.e. for db fields I usually name fields: tablename_fieldname cuts a lot of ambiguity)

Check your application file structure, everything in one directory works whan starting out but as you increase your skills/apps/libraries you will want to organize your project directories more logically. Looking at some FOSS apps and how they organize things can help.

Anything you feel weird about - research look at other examples, stack overflow results usually gives you a variety of examples you can usually spot better methods from bad ones.

Help out others a little on forums, get shot down for your crazy methods, figure out why (done this a couple times, stings but it helped, some pros only respond to people posting bad answers)

And as other said, write, write, write, the more you write the more you will get a sense you are spinning your wheels on something or some part could be done better.

- Algorithms and Data Structures

- Operating Systems

There are more but it helps to at least learn the fundamentals of these two. If you feel up to it, there are these too:

- Computer Organisation

- Compilers

There are more which you could look at depending on your interests: Computer Networks, Computer Graphics, Databases, Computer Architecture, Programming Languages, Theory of Computation, Distributed Systems

Many of these can be learnt from a good online course or a good textbook

Besides these, I found it very helpful to try and contributed to big free software projects. The standards are high and you'll get a lot of feedback on your patches. If you pick a project like something by Mozilla, you'll also have bugs/tasks specifically earmarked for people who are new contributors.

I forgot to add: Reading good code really helps. However, for this, I found that the most effective way was to try and fix a bug or add a feature which leads again to my suggestion above. This way, you have a 'purpose' with which you read code and it helps navigate a large codebase.

Good luck!

I know way more (maybe 10x more) tools, more approaches, and more languages, than if I'd just stuck with self-taught approaches like books and tutorials. Those have some utility, but it's a very distant second.

Plus, actual people help you get better jobs (or investors) even years after you part ways with them.

In short, best practices require practices, not theory.