Me postulé a través de una recomendación de un empleado. El proceso tomó 3 semanas. Acudí a una entrevista en Rokt (Sídney) en jun 2025
Entrevista
My experience with Rokt has been highly responsive, accommodating with interview arrangements, and proactive in providing feedback to my questions.
Overall, I had a very positive experience with the people I met, both online and onsite. I especially appreciated the fast-paced process—most of the time I received my results the day after each interview, along with scheduling for the next round.
Another highlight was that after completing a few rounds, the HR team arranged a dedicated meeting to walk me through feedback from each interviewer. This level of transparency was a huge plus, and the feedback I received throughout was fair and constructive, adding to the overall positive experience.
In the end, I chose an offer from another company based purely on my own career direction at the time. However, I truly appreciated the tremendous effort the Rokt HR team made to persuade me, which left me feeling respected, valued, and fully supported.
Preguntas de entrevista [1]
Pregunta 1
In the Bar-Raiser Interview, I was asked: What do you think of the transparent compensation policy that Rokt applied?
The interviewer at Rokt was very friendly, making the process relaxed and comfortable. The interview lasted about an hour, focusing on technical questions. Overall, the atmosphere was welcoming and pleasant.
Me postulé en línea. Acudí a una entrevista en Rokt (Sídney) en jul 2025
Entrevista
The interview process included a technical round focused on a machine learning problem. The specific task was to design and discuss a recommendation system. Be prepared to talk about different approaches and how you would evaluate the model.
Preguntas de entrevista [1]
Pregunta 1
The main technical question was a deep dive into the architecture of two-tower models. They specifically asked how I would design and implement one for a large-scale recommendation system, focusing on how to build the user and item towers separately.
Questions about oa and vi can be found in the related websites. ML coding: high-dimensional data with naive interviewers, pre-process, model, evaluations. More exercises are important. No depth there, I believe the staff there still use the oldest techniques, and there is no need to hope them to dive deep into the SOTA techniques and solutions
