One phone interview, one take-home project to do (think level 2/5 Kaggle competition style with a low acceptance bar), a technical interview, then had to present a paper. Got rejected after this step.
The people individually were all nice, from technical people to the very verbose and transparent HR. However, it doesn't wash away the fact that ML6, like any other tech company out there, makes you jump through hoops during an efficient and highly random process.
The take-home project was easy and outdated to the point that you needed to fix a couple of bugs in their boilerplate code. The technical interview was fine. Then I was sent a paper and was asked to prepare a presentation. It was explicitly specified that the audience would be ML Engineers and other technical staff. Naturally, I gauged the presentation for such audience, with an overview of the math, pros and cons of the paper, and finally putting it into the bigger context of other methods that do something similar to the paper's subject, review available tools/packages, etc. During the paper interview, after my presentation I was asked several technical (light, not in depth) and non-technical questions (mostly how'd you describe this to a client, etc.) which I tried to answer. It seemed like the people present had not read the paper themselves, or at least not in depth. A few days later, I was notified by their HR that I was rejected because they thought I was not skilled enough to deal with clients directly. I do not accept their conclusion and have a job history and a new job - all client facing - to prove it, but they are entitled to their opinion and it's fine.
The problem is, if they want you for a client facing position, wouldn't it be logical to measure your soft skills early on before making you spend several days of your time coding and preparing a presentation? Second, their interview process was not aligned with their expectation from it. I have a graduate degree in ML and years of experience, so I am used to adjust the level of my presentation according to my audience, and experienced at how to prepare for each audience. The paper presentation task I received felt more like either a trick or a careless mistake in hindsight; if they wanted to asses my soft skills they could have been upfront about it and then I'd have prepared a different presentation and pitch, and prepared myself for their questions (that is unless at ML6 office, the client just walks through the door and sits at your desk and demands you explain in kindergarten terms how Gluon works or something!). Last, I have years of experience in different tech positions which required dealing with a wide range of internal and external stakeholders. To then be rejected on one of the strong points of your career through such a haphazard process, imo, is unacceptable. The whole thing feels more like a scientist/engineer-adjacent person's idea of a clever interview process.
TL;DR, imo, is that they are much nicer than your average tech company (Meta interview was hell!), but still their process if flawed and leads to a waste of time of the applicants, which is ABSOLUTELY unacceptable. So, my advice to anyone applying for ML6 is, ask pointed questions about each task, try to pinpoint what is the expected outcome of each meeting/interview and which of your skills they are assessing in each step.
Wishing you all better luck than I had with ML6.