I was contacted by a recruiter about a Senior Solutions Architect position where they wanted a search expert, and the role sounded like a perfect fit for me. I took an hour long assessment including a personality test, a lengthy math test, and a vocabulary test. I then met with the hiring manager, and someone in a similar position for a technical interview. They blocked out an hour for the technical interview but the interviewer ended it after 20 minutes.
I was very surprised when I was turned down after the technical interview for not having enough experience around vector search and RAG since none of the interviewers asked a single question about these topics. In fact no one asked me any tech questions at all during the interview process aside from asking me to describe my experience.
I contacted the recruiter and complained that my interview was cut short and no questions were asked about the subjects that they said I had no experience in. To their credit, they then scheduled another interview with a more senior engineer in the department I was interviewing for, but we only had 30 minutes to talk so this interview was very rushed and went over the time allotted. This interviewer only asked very complicated questions about AI, which is difficult to discuss in a rushed interview because not all companies are using the same terms to describe parts of the process as the technology is very new. I had actually watched some youtube videos beforehand to check out how they were referring to different parts, but it didn't really help since the videos used slightly different terms depending on who at MongoDB was describing it.
This time they said I did know vector search and RAG, but they turned me down because I didn't have enough experience creating LLMs and text embeddings. I was the main person troubleshooting text embeddings after other engineers created them at my last position, and I have created other types of gen AI models and used several different LLMs. The job description only asked for 3 years of experience with searching databases and software development and no AI experience, so it felt like the goal posts were constantly shifting during the interview process and they weren't sure what they wanted. I have almost 20 years of experience in tech and worked on vector search 5 years ago for a few years, and it was just baffling that somehow this wasn't enough. Especially how they threw in the requirement that I needed to have experience creating LLMs since that's not part of the job at all and not something I'd need to do. The job description said that one of the perks of the position was that you'd get to learn about vector search and it didn't mention needing any knowledge on RAG, building LLMs, or embeddings, but by the time I was interviewed it was a completely different position.
All of the questions during the first tech interview were "What is your experience with scripting/databases/AI/Networking/Security/etc?" with no technical or follow up questions asked to test the extent of my knowledge. This would be an extremely easy interview process to get through if you were lying about your experience.
The second tech interview was extremely complicated and he asked a lot of questions where I had to constantly consider my past NDAs. For example specific information on how my past company was using vector search, such as how they were processing embeddings and their exact specs such as how many dimensions they were using. For this he didn't seem to be referring to how many vectors were in the dense vector array and seemed to be using the term to refer to something else, but I wasn't really sure what it could be. He also asked a lot of questions that had nothing to do with the job at all, such as how I was using Stable Diffusion (an image AI), and at times it felt like he was just looking for advice on his personal AI projects. I wanted to explain to him how he could fix his home setup once he told me his specs, but then I started to question what this had to do with the position. None of the questions asked had anything to do with the job description.
I did think it was funny that the first interviewer told me that he had hours of extra time at work and didn't do a lot day to day and the second interviewer said he was always busy and on the phone and they really need more people, even though they supposedly have the exact same position and expertise at the company. Instead of having people do two management interviews (the next interview would have been with the VP, and then a presentation) they should really let candidates do two tech interviews so the interview is long enough to cover the basics/search and AI. That way we can also meet more members of the team as this difference in descriptions was a red flag for me and made me lose a lot of interest in the position.
Also, the pay band is very low and not competitive for the amount of knowledge they expect you to have on AI beforehand.
