It started with an introduction that lasted like less than a minute, and straight away the interviewer shared a link where we went to solve a Hard Problem i.e Edit Distance with little to no interaction with the interviewer. Afterwards, there was a bit of baiting around i.e tell me which Classic ML model you like and explain that model, once I explained that it would be a Decision Tree family, then they started asking what would be the "loss" function and more questions that were purely adverserial attacks which is fair in this context but after having created a sub optimal pseudo code for Edit Distance for the last 30 minutes, this was doomed to fail, at least for me.