650+ LeetCode Problems Later: What Actually Changed

What consistent DSA practice actually improved in my day-to-day engineering work, and what it didn't.

Somewhere past the 400-problem mark I stopped tracking the count for the sake of the count. Here's an honest breakdown of what 650+ problems on LeetCode actually changed in how I write code, and what it turned out not to matter for.

What got better

Pattern recognition. Sliding window, two pointers, monotonic stack — after enough reps, a class of problems announces itself in the first sentence of the prompt.
Reading code under time pressure. Contests force you to parse someone else's constraints fast. That transfers directly to reading a PR.
Complexity intuition. I now have a rough cost estimate in my head before I write the first line, instead of after profiling.

What it didn't fix

Grinding problems doesn't teach you how to structure a real codebase, how to name things so a teammate understands them in six months, or how to debug a production incident at 2am. Those come from shipping software, not from solving isolated puzzles.

The actual takeaway

Use DSA practice as a sharpening tool, not a substitute for building things. The two compound — algorithmic fluency makes the build-things part faster, not the other way around.

two-pointer.ts

function isPalindrome(s: string): boolean {
  let left = 0
  let right = s.length - 1
 
  while (left < right) {
    if (s[left] !== s[right]) return false
    left++
    right--
  }
 
  return true
}