The DSA Shift
Hot take: I think software engineering interviews are measuring the wrong thing.
For years, the path has been pretty predictable:
Learn DSA → grind LeetCode → solve 500 problems → prepare for interviews.
And honestly, DSA is valuable. It teaches problem-solving and builds strong fundamentals.
But something has changed.
AI can now write a lot of the code.
It can implement algorithms, optimize functions, explain solutions, and get you surprisingly close to a working implementation in seconds.
So what becomes more valuable?
Knowing what to build, why to build it, and how to build it efficiently.
That's where I think System Design deserves much more attention.
Give two engineers the same AI coding assistant and ask them to build a URL shortener.
One might immediately start asking AI to write the code.
The other might first ask:
How much traffic are we expecting?
Do we need caching?
What database makes sense?
What happens if a service goes down?
How do we scale this from 10K to 10M users.
Both might eventually write working code.
But the second engineer is thinking beyond the code.
That's engineering thinking.
And yet, when we talk about practicing engineering skills, the names that immediately come up are:
LeetCode. GFG. Codeforces.
But what about System Design?
Where do you actually go to practice it consistently?
Where can you solve real-world architecture problems, understand trade-offs, get better over time, and maybe even build a System Design streak?
Imagine going from designing a URL shortener today to a chat system next week, a food delivery platform after that, and eventually tackling genuinely complex distributed systems.
Not memorizing architectures for an interview.
Actually learning how to think like an engineer.
I genuinely feel there is a gap here.
DSA shouldn't disappear. It's still important.
But maybe it's time we stop treating it as the ultimate proxy for engineering ability—especially for engineers who already have a few years of experience.
Code can increasingly be generated.
Engineering decisions can't be outsourced that easily.
So I'm curious:
If you wanted to seriously practice System Design today, what platform would you use?
And if there isn't an obvious answer, what do you think that platform should look like?
#SystemDesign #SoftwareEngineering #TechHiring #ArtificialIntelligence #FutureOfWork