I had three in person interviews totalling about 6 hours; 2 in one four hour session and a following one on a different day, and a take home task. All took about 3 weeks, was told the take home was the final step and then heard nothing for more than 2 weeks. Finally managed to chase them up (they like communication to go solely through the recruiter which made this a pain) and got sent a rejection email with no details, despite having a discussion in my first interview (I was trying to sound out if there would be feedback if I got to the take home stage) about how unfair it is on a candidate to have them put in so much effort to an application, particularly a take home task, and give them no feedback. Sent a follow up requesting feedback which was ignored.
I think the in person aspect is a really good thing as it makes communication so much easier, but it is a big commitment for the candidate to travel to their offices multiple times; they are very accommodating of your time by offering to do interviews outside of working hours, but the idea of taking evenings and weekends for such long interviews is not that appealing. The interviews themselves were quite repetitive on the technical side, all three were more or less the same sort of standard quant questions (to the extent that I had some questions repeated just with different numbers) so the long slog felt pretty unnecessary. They seem to value the culture fit, and so want multiple people to meet the candidate (everyone did seem pretty nice) but it felt overlong, particularly if you get nervous in interviews it's just so draining to be bombarded with questions for 4 hours straight that I think it stops being productive. I was surprised that there was very little emphasis on modelling or coding, just really basic stuff, and more generally I tried really hard to get a sense of what the day to day is like (presumably they're not actually solving brain teasers) and couldn't get any sort of detail beyond really generic stuff (giving an example of the process is surely not leaking IP!)
I was told that the task home would give a good idea of what the work looks like, and found it pretty strange and disappointing. It was a standard data science task but very prescriptive in how you should approach it, and seemed to leave little room for experimentation but was more a turn-the-handle exercise. There is even a short call with them to check in how it's going and make sure you're not going down the wrong path, where they doubled down on how you should/must approach it. I really didn't understand the point of that call because I was asked what my first steps had been and plan was, but was given absolutely no feedback, there was no discussion at all, just noncommittal noises from them (I hope that's not reflective of their process!) I feel I must have missed something - this is why feedback would be so useful!
