The biggest change I've noticed is that AI has shifted the bottleneck from writing code to evaluating it. Generating a solution is easy now; knowing whether it's correct, scalable, and maintainable is where engineering judgment really shows.
Working on AI-driven projects at IT Path Solutions has reinforced that the strongest developers aren't the ones who avoid AI they're the ones who question its output, catch edge cases, and make informed architectural decisions. LeetCode-style problem solving still builds that foundation; it's just no longer the whole picture.