I asked for a referral. I wanted a position related to Machine Learning, but I don't have any work experience besides academic work. My friend assured me that it would be fine.

There were two telephonic interviews and one onsite interview. However, the process felt messed up. It felt like they interviewed me for a non-ML role and then rejected me for the ML role.

Phone 1: Spent 20-25 minutes out of 45 minutes talking about what I work on right now, since they were doing something very similar. Then coding – something related to an interval problem.

Phone 2: Same thing, spent 25 minutes out of 45 minutes talking about what I work on right now, and then a 2 Sum problem!

So, my impression was that I was interviewing for a Machine Learning role. In between the phone and onsite interviews, the recruiter changed. The first recruiter told me there would be a Machine Learning round for the onsite. The second recruiter said, "We can't tell you much, but yeah, there will be ML questions." The procedure isn't standardized. She mentioned before the interview that she would tell me about all the rounds and what they would be for the onsite, including giving hints about the ML questions.

Totally different things happened at the onsite:

Onsite 1: Again interested in what I work on. Asked me what I would prefer: model building versus building an ML pipeline. (And THAT, according to them, is an ML question!) And write a snake game. Easy.

Onsite 2: Design. "An autonomous car collects data. We want to get the data out (say, speed/location, etc.) and also push data (like some commands). How would you design?" A broad question, but whatever we could do.

Onsite 3 (30 min): Lunch.

Onsite 4: Behavioral. I believe this judges your level as well as whether you're a jerk. Questions like: "Have you faced a situation where your design was wrong?" etc. Don't remember.

Onsite 5: Ops. Weird, right? For Machine Learning, they asked me about Ops!? If you're an Amazonian, you would nail it. How and where would you put metrics?

Onsite 6 (30 min): Car ride.

Onsite 7: Manager round. Something like behavioral.

After the interview: Recruiter: "Sorry I couldn't give you details about what all the interview would be. But did you find anything odd?" Me: "Yeah, there were not ML questions. And Ops? I don't know." She: "Oh, okay. How did you like the team...?" Me: ...

Result: Recruiter: "Well, we're not moving forward because your ML experience isn't meeting the bar." I was like, "WTF? You didn't ask me any ML questions." Me: "Could you at least give me feedback?" She: "First round, you were asked ML questions, and they were not satisfactory. You don't have work experience in ML." Me: "Well, no, there were no questions. And yes, you're right, I have academic experience." (I was referred, and the person had made sure it was fine.) She: "You were more of a generic role, but we only have 1 senior engineer role open, and you're not meeting the senior level." Me (still): "Without asking ML questions, how did they decide it's a 'no'?"

I believe what they interviewed for was a completely different role and a senior role – like, 'let's give it a shot.'

I'm baffled by their interview process. If you take an ML interview and reject me because I didn't do well on questions, sure, that makes sense. But why would you reject for something that you haven't interviewed me for?