Voyado interview questions

The interview process included a take-home GenAI/agentic systems assignment followed by a technical discussion. The assignment itself was interesting and relevant to modern LLM engineering. The HR was so kind and helpful. However, I felt some of the feedback during the review process was overly high-level and in several cases inaccurate, which gave me the impression that the implementation and documentation may not have been reviewed deeply enough, despite the explanations provided during the interview and in the README file. For example, some concerns raised about orchestration order, model usage, and guardrails were already addressed in the codebase and documentation. In one case, an example prompt mentioned as bypassing the guardrails was successfully blocked when I tested it again afterward. In another case, feedback was given that the system used only one model for all purposes, despite the implementation and presentation clearly explaining that different models were being used for different responsibilities. I also felt the feedback team was significantly stronger in traditional software engineering than in modern AI/LLM engineering topics. During the session, I showed and explained how I had implemented topics such as LLMOps, observability, prompt engineering, evaluation workflows, and modern GenAI orchestration patterns inside my own system, and I also discussed how these concepts could be implemented in production environments. It was quite clear to me during the discussion that many of these topics and tools were new to them and I genuinely felt like I was participating in a training session for the team rather than an evaluation interview. Because of that, the rejection afterward gave me the uncomfortable feeling that my GenAI knowledge and preparation may have been used more for educating the team than for fairly evaluating my candidacy.