The process began with an initial discussion with the recruiter, which seemed to go well.

An interview was scheduled. Although the interview was rescheduled multiple times throughout the process, this was acceptable.

The interview itself was an LLM quiz. I was provided with multiple resources to prepare, however, they barely mattered. For instance, I received no questions regarding LLM architecture.

The interviewer did not seem very confident with the topics discussed and was heavily biased towards certain solutions without good reason.

I was asked about the trade-offs between RAG and fine-tuning for a chatbot answering questions on a broad domain. The interviewer heavily insisted that RAG was the right choice (?), and I still struggle to understand why, as there is room for both.

It seems as though one is supposed to guess what they prefer as an answer, rather than arguing what is actually true.

In the end, I asked for feedback but received no clear insight into what I answered incorrectly during the interview. I was only given some subjective insights.

Explain LORA fine-tuning.