Matroid Deep Learning Engineer interview questions

This was a while ago (in 2019). The question was a simple question finetune a tf classifier for a specific task. However there were glitches in the weights shared (it was also a Caffe Model - who in academia ever touched this antiquated language) and I was given a very short time to work on the project. After spending 2 sleepless nights and getting back to them - they didn't even have the courtesy to schedule an initial interview. It is okay to reject but respect the other person's time and effort they put into the interview. It says a lot about the type of company you want to become.

Given a deep learning challenge, finish it within three days. The task itself isn't super difficult, but still takes decent time to complete. There're four steps listed in the description required to accomplish. However, the interviewer in the end judged the performance based on things never mentioned in the description. They'd expected you spending lots of time finishing the first round project and also guess what they liked you to do. In my opinion, it's a total waste of time and irrespective to the candidates.