I had a HR phone screen, a one hour light technical covering basic knowledge of stats and machine learning, followed by a take-home project.
The phone screen and light technical were standard and fine.
The take-home stunk of trying to steal free work. And if it wasn't meant for that, it certainly felt like it. I was given an arbitrary amount of time (they asked me to tell them when I would submit the project for scheduling review), and a quoted "average completion time" of 3-4 hours. That was a joke. The project was described as a "case study" that came with a ~1gb dataset in two tables. The ostensible goal was to calculate some grouped metrics with some instructions -- "How often does X happen in a group of actors with condition Y". Had this been the actual project, it might have been reasonable. However, it was actually an ask for a fully extensible framework, with loggin, debugging, analysis, arbitrary adaptation to new input parameters (of which no examples were given other than the case study at hand), and complete terabyte+ size scalability. You know, the sort of thing that would a core system at a company like this.
After putting in 5x the amount of time "as average" (since I had never touched their data before and have no framework available from having done these before), I received a rejection email. This is despite having been assured I would be presenting my results and discussing them with the relevant parties.
To their credit, they gave feedback. That said, it was telling. Amongst the feedback was a complaint that my variable names were too specific to the _case study they asked me to do_, as if I couldn't trivially replace the variable names.
I can not be sure they were trying to steal my time and my work, but I will reiterate that at minimum, it feels that way. This was an unreasonable project to give as a take-home, and I do not recommend entering this process.
