J'ai postulé en ligne. J'ai passé un entretien chez Criteo (Paris) en avr. 2026
Entretien
Around 5 rounds of interview, tough but fair questions, candidates are usually either in a deep learning or a strict machine learning track. Tests both coding and machine learning, the machine learning track is more mathematical have a good understanding of optimization theory, maximum likelihood etc. Coding exercises also have a data science slant to them.
Autres retours d’entretien d’embauche pour un poste comme Machine Learning Engineer chez Criteo
J'ai postulé en ligne. Le processus a pris 1 semaine. J'ai passé un entretien chez Criteo (Paris) en janv. 2021
Entretien
Responsable ressources humaines : call de 30 min , test en ligne : QCM : langage de programmation au choix: pour mon cas python , entretien technique : 45 min code python 45 min machine learning/ deep learning : pas trop compliqué
Questions d'entretien [1]
Question 1
-Regression logistique
-Random forest
-Variable alétoire , donnez des exemples
J'ai postulé via un recruteur. J'ai passé un entretien chez Criteo (Paris) en avr. 2017
Entretien
The hiring process imitates the style of SV companies, most notably that of Google. Before the first interview, I received a long pdf document containing a detailed formal description of the entire interview process. The whole thing reminded of a bureaucratized procedure in a public service and the described process seemed ridiculously tedious next to the salary range they offered (according to glassdoor).
In the technical interview, the interviewer was unsympathethic and borderline rude, definitely not the kind of person you'd like to be working with. He had a very strong French accent and I was having a hard time understanding him. Whenever I told him I didn't understand what he was asking, he would get irritated. Within the first 5 minutes I understood that he only had superficial knowledge of the domain he was supposed to be an expert in, and he couldn't really understand the things I was telling him (he had never heard of concepts such as time-series forecasting and regularization). When the conversation went to maths, stats, and optimization at the heart of machine learning, I could tell that he couldn't follow and would try to change the subject.
He tried to "corner" me a lot with abstract questions and kept manipulating the conversation looking for ways to confuse me and make it seem as if I couldn't answer the question (while in actuality I simply couldn't understand his English). I had a hard time answering his logic puzzle as a result of his inability to describe the problem in correct terminology (he didn't really understand probability distributions, he kept misusing the word "recursion"). I did come up with the right answer in the end, but gave me a negative feedback anyway for "having a difficulty understanding his questions". I felt I was basically penalized for not understanding broken English and for being overall more knowledgeable than the interviewer on the topic I was being evaluated.
Overall, most of the interview felt as if the interviewer was trying to match his knowledge against my own more than anything else. When asked, he did not know what position I was interviewing for (which I found absurd) and at that point I realized that I had been wasting my time.
Questions d'entretien [1]
Question 1
Questions on logic and basic probability, describe a machine learning algorithm from scratch.