Recently Meta announced that they will allow the use of AI tools in interviews. This changes the way candidates prepare for these interview. After having solved 700+ Leetcode problems I realized that after a certain point, the learning saturates.
There is also very little you can learn from grinding coding websites that can be translated to real-world SDE tasks. Sure, knowing how to work with heaps is a good skill to know when you want to implement small in-memory priority queue or knowing topological sorting in case you are trying to make a package manager. But honestly, it doesn’t take more than 1 day to learn any of these concepts, also with hundreds of libraries available that let you simply call the required functions, you don’t need to spend days trying to master these concepts.
To be really good generalist software engineer, you need to master The Black Box Method. You don’t need to know how push and pop algorithm for heap works under the hood, but you can simply understand what heap is and how pushing and popping items in heap can help you get items faster in O(log(n)) time. Look the type of problems that you can solve with heap. And when in future you see a probelm your brain will automatically tell you that this is a heap problem.
Cool, but how is this related to software or production engineering interviews?
In real world, nobody cares if you Google an error or simply copy paste the error trace to ChatGPT. What’s important is to solve the problem in front of you. Because of the growing supply of software developers and with rise of LLMs the old DSA+Algo style interviews are becoming useless. Its becoming easier and easier to game or cheat the system. Just solve the top 15 most frequently asked questions for a company and there is a good chance that you will clear the interview.
This is the main reason I created sttrace.com(yes, the name in inspired by linux tool strace, that domain name was costly🥲), having realized that real world software engineering requires you to be decent in a wide range of skills which are very hard to develop if you haven’t faced enough problems or production outages.
It has a wide variety of problems focusing on Debugging, DevOps, SRE, and Security. The good thing is, you can solve the problem however you want. You are not forced to use a certain tech stack or algorithm to solve a problem, the submissions are evaluated based on the final results. I add new problems daily, and I’ll soon be launching an internal job board.
As AI changes the interview landscape, the engineers who thrive won’t be the ones who memorized algorithms, but the ones who’ve practiced solving unpredictable and messy problems. And that’s the vision behind sttrace.com
Top comments (0)