- The discussion started with the presentation of the departments. There are 100 employees in Romania.
- Then for 10-15 minutes I presented the projects I worked on, emphasizing the ML area. What articles I wrote, what models I implemented, how I implemented them, how we organized ourselves on the ML projects.
- Then followed a series of common sense questions in ML, nothing difficult: what do you do if you have over-fitting, what is a logistic regression
- What is the difference between a sigmoid and a softmax. Give examples of other activations
- If you were to make a classification model, how would we proceed?
- What technologies have you worked with Keras/Pytorch and in what language
- At the end of 15 minutes, a coding problem, with the mention that you can write it in any language (even pseudocode), where you want, even a notepad. I had to pretend that I had an image and translate it to the right. I wrote in python on leetcode and I also run it to test, everything was good.
I did not receive the offer. Although I answered all common sense questions and solved the problem,
a few days later, I received an email saying that they are satisfied with the concepts I have and the projects I worked on.
But unfortunately, they are looking for advanced C++ people, not Python.
During the interview, I had no C++ questions. Just one question, which language do I feel more comfortable with, the answer being Python because in this language I have been preparing for a similar position in recent years.
My opinion is that you should apply if you want a similar post outside. If they insist on such a position with knowledge of C++ (which they have, but have not been evaluated), there is a great possibility that you will not evolve on the ML side in the company and work on other tasks, probably embedded.