What is Your Approach to Solving Problems With Code?

This may not work for coding challenges and Im not sure why number of lines matter. But below is what I follow for real time problems

1. Never start to solve the problem directly from code.
2. Have a pen and paper, visualize what exactly you are trying to solve. Think twice, code once

A key for me is reduction of variables. So like removing lines of code debugging etc.

I think this applies to general problem solving. Figure out what you don't have to worry about first.

A good teacher on youtube once said:

1. Understand the problem
2. Think of a solution
3. Implement that solution with brute force method. That is, just get a working prototype.
4. Then look at your code again, see where you can optimise.

This approach really worked for me!

I tend to find that if a Developer is going to go "off the reservation" they'll do it in chasing that bit of optimization that gets us from 6 lines to 5 or whatever... so I try to stay focused on the words "Solving Problems" rather than "with Code".

First iteration: Solve the problem.

Second: Clean up egregious stuff.

Try to stop there unless new information comes to light or unless the system is life-and-death - because otherwise, you can spend a lifetime tweaking something - the returns diminish rapidly.

I have learnt the hard way that diving headfirst into code is a bad idea.
My process is simple.

1. Understanding the problem - You can research with google or AI, YOUR CHOICE.
2. Plan the solution -You could use diagrams or just write out the steps you think are needed in plain English.
3. Turn your plan into something that looks like code but with 0 syntax errors, pseudo code.
4. From there, you are free to use any programming language you want to build whatever it is you are building.

This reduces time spent figuring out how stuff work when writing the actual code.

My first step to solve a problem, is setup the state structure which collect a necessary data from input and good enough to represent the goal data.
After that step I just need to make two method:

• convert input to state.
• find out the result by use state

This method work even complex problem, when state are some stream.

I usually start by breaking the problem into smaller, manageable parts to clearly understand what’s being asked. Once I’ve grasped the requirements, I try to relate it to any standard approach or technique I’ve learned—either from coursework or practice. I actively solve daily problems, like LeetCode’s Problem of the Day (POTD), and regularly follow structured problem-solving sheets like Striver’s, which also include detailed solution approaches. This routine helps me recognize patterns more quickly and apply the most efficient solutions. Hope this works!
Just break the problem into smaller problem -- (just like tagline of recursion 😅)

My approach to solving problems with code starts with understanding the why—what exactly needs to be solved and who it impacts. I break the problem down into smaller, manageable parts, then plan a solution before touching any code. I usually write pseudocode or draw diagrams to map things out, then start building iteratively—testing and refining as I go. I also try to keep things simple and readable, because future me (or someone else) will thank me later. And when I get stuck, I ask questions, read docs, or look for patterns in how others have solved similar issues. Problem-solving with code is as much about mindset as it is about syntax.

Back in the day, I used to do codeforces. good times.

I totally agree. Problem-solving is the real core of programming. My usual approach is to break the problem into smaller chunks, write down what I think should happen step by step, and only then start coding. I also try to rewrite the problem in plain language before jumping into syntax.

Funny enough, I sometimes take short breaks when stuck, like browsing random things to reset my brain. The other day I ended up scrolling through the kfc menu while debugging a tricky function 😅. It worked, though the solution clicked right after!