after getting rejected a bunch, i started asking recruiters for feedback. most gave generic "we decided to move forward with other candidates" but a few were honest. combined that with self-reflection after each round.
here's what i found after 23 interviews:
the 4 failure modes (and how often each killed me):
| Failure mode | How often | What it looks like |
|---|---|---|
| Didn't recognize the pattern | 35% | you stare at the problem for 10 min, try random things, maybe brute force it but can't optimize. interviewer gives hints you don't understand. |
| Recognized it but too slow | 30% | you know it's DP or BFS but spend 25 min coding, run out of time for follow-ups. the "easy" first question eats your entire slot. |
| Code works but can't explain tradeoffs | 20% | you solve it but when asked "why this over that?" or "what's the space complexity?" you fumble. interviewer thinks you memorized the solution. |
| Communication failure | 15% | you solve it silently, or jump to code without discussing approach first. interviewer can't follow your thinking so can't give partial credit. |
the fix for each:
pattern recognition (35%): this is a prep problem, not an intelligence problem. if you haven't seen the pattern before, you can't "figure it out" in 5 minutes under pressure. the fix is covering all 12-15 core patterns with 5-6 problems each. after that, every new problem maps to something you've seen.
speed (30%): stop solving problems without a timer. set 25 min for mediums, 15 for easies. if you can't finish in time, your implementation is too slow — you're probably writing code before fully thinking through the approach. the fix: spend 5 min on approach (talk it out), THEN code. counterintuitively, this makes you faster because you don't backtrack.
tradeoff knowledge (20%): after solving any problem, ask yourself: what's the time/space? could i do better on space? what if the input doesn't fit in memory? what if i need it sorted? these 4 questions cover 90% of follow-ups.
communication (15%): practice talking out loud while solving. literally narrate: "i think this is a sliding window because we need a contiguous subarray... let me verify with the example... yes, my window expands when X and shrinks when Y." sounds awkward at first but interviewers love it because they can give you hints if you're going the wrong direction.
my results after fixing each one:
- before: 4/23 passed (17% pass rate)
- after 3 weeks of targeted fixes: 5/7 passed (71% pass rate)
same person, same brain, different approach.