How is dev hiring actually being done in startups & enterprises in 2026? I will not promote

The traditional technical screening is somewhat dead now . LeetCode/ OAs - Mostly affected by ai or memorization. It tells you nothing about whether a candidate can handle production systems. Take Homes - Anyone can feed the problem statement into Claude or Chatgpt, let it generate the codebase , tests and a clean README , and hand in without understanding the internals. In day-to-day engineering right now , I think the sweet spot is the duo of "Vibe coding" / AI velocity: Using agents and LLMs to scaffold, and build at 5x speed. Deep engineering fundamentals: Knowing how the underlying system works so you can actually debug subtle race conditions, diagnose memory leaks, read stack traces, and verify when the AI hallucinates. Folks actively involved in hiring how are startups or companies are currently testing this combination?? Is there any tool or platform out there for this?? Curious what current hiring pipelines look like that aren't just burning dozens of engineering hours filtering out copy-pasters.