Me postulé en línea. Acudí a una entrevista en Rakuten
Entrevista
Upon arriving for my interview, I was informed that HR had forgotten to include the interviewer in the meeting invite. They attributed this to a new HR system implemented six months earlier. While the issue was acknowledged, it impacted the overall experience.
El proceso tomó 2 meses. Acudí a una entrevista en Rakuten
Entrevista
I had a first screening interview after ~3 weeks, and waited 4 weeks to pass the technical interview which has 2 sessions (each session 1h with a staff member). The way to schedule the interview was particularly not smooth and impractical. I had to exchange by emails to fix the dates and get confirmed.
Preguntas de entrevista [1]
Pregunta 1
- Implement cross entropy without using defined function in torch
- Implement a data loader of large JSON text database in pytorch
- Questions about: difference between SFO and RL(HF), what consumes more of GPUs during the training of LLMs, and many other questions about parallel training
The interviewers seemed not too knowledgeable about their questions and likely asked an LLM to generate them. The answer that we discussed was divergent and did not convince me. Both interviewers were very unfriendly and not collaborative during the interview.
Long interview process, everything is automated by AI, so no HR to communicate with.
First round with hiring manager - technical discussions, and design choices.
Second round - coding in python loss function, linked list, sampling function, very algorithm related. Bootstraping techniques
Final round - discussion with hiring manager and director
