Our company recently advertised a junior developer position on LinkedIn for frontend, backend, and QA roles with compensation reaching ₹20L. We received over 12,000 applications but had to screen out more than 10,000 right away due to mismatched skills or irrelevant experience. We did this to save everyone’s time rather than putting unqualified people through lengthy interview processes.
During our technical interviews, we tested fundamental programming concepts and data structure algorithms including binary trees, priority queues, linked structures, breadth-first search, depth-first search, and similar topics. We even permitted applicants to use ChatGPT for coding solutions. But when we asked them to explain time complexity, space complexity, or walk through their code logic, most couldn’t provide coherent answers.
Many applicants seem to be doing what I call ‘blind coding’ where they just copy and paste AI-generated solutions without comprehending how the code actually works. This trend makes it incredibly challenging to identify developers who genuinely grasp their own implementations.
We’re wondering if our interview approach needs adjustment, or if junior developers today need to focus more on truly learning code fundamentals instead of relying on AI tools without understanding.
450 interviews is way too much. Sounds like your pay ain’t matching what you’re testing for - ₹20L pulls in senior guys, but you’re askin’ junior-level stuff. The ChatGPT thing is weird too - lettin’ them use it then askin’ them to explain feels like a trap. Either ban AI or write questions that fit it. Try pair programming instead - you’ll see how they really code.
Wait, you let them use ChatGPT but then act surprised when they can’t explain their logic? That’s backwards. Try giving them working code to modify instead - like ‘this function works but needs to handle null values, how do you change it?’ You’ll quickly see if they actually get what’s happening or if they’re just copy-pasting.
You’re seeing a real shift in how people learn programming now. AI tools made coding more accessible, but they’ve also created developers who can get code working without really understanding what’s happening underneath. I’d change your interview approach. Skip the complex algorithm coding and focus on how they think through problems instead. Give them scenarios where they pick between different solutions and explain why. Ask stuff like ‘When would you use a hash table vs a binary search tree?’ or have them debug code with logic bugs. This shows you who actually gets it and mirrors real work better. Plenty of solid junior devs bomb the textbook algorithm tests but are great at solving practical problems and picking things up quickly.