I did a recruiter screen that was followed by a coding test. The coding test was surprisingly hard, and while I didn't at the time I believe the expectation was that I would use AI to finish the coding exam. That being said, the exam was incredibly difficult and one of the questions took a friend who is a developer 20 minutes to refactor the answer that AI gave him for the problem prompt. There were 13 questions, about half of them were about language fundamentals and the rest were code. For example, one of the questions provided a quadrant filled with different plotted points bounded by an arc and the goal of the question was to use the plot points to calculate pi. For a solutions architect role this is absurd and a red flag in my mind