The interview process consisted of two rounds: one virtual and one on-site.

The virtual interview focused on my personal project in Generative AI. Although I was initially meant to complete a LeetCode-style question, the interviewer, having reviewed my project's code, decided against it.

The final on-site round was an intensive 1 hour and 30 minutes. I was thoroughly questioned on the foundations of LLMs, their architecture, fine-tuning, and related topics.

According to the feedback received, I had no drawbacks. The company ultimately decided to proceed with internal candidates who possessed existing knowledge of RASA for their chatbot development.

1. If you had a limit on your paid LLM and you run out, how can you mitigate that?

2. Talk about the architecture of the Transformer.