I just received a take-home exercise, and I’m completely bored by it. They didn’t even ask if it was a good time, just sent a link with a one-week deadline.
It’s one of those tasks with a gig of random data, full of nested fields, and no clear instructions. I’ve spent most of my time ranting about how I shouldn’t have to do this, thinking I have better things to do than figure out some company’s messy schema. Over the years, I’ve grown to dislike wrangling data with pandas.
The problem is, I really need a job, and these are the only gigs I get responses from. I know I should be grateful to even get a reply, considering how competitive the market is.
To wrap this up with a question: How much time do you all actually spend on these kinds of tasks? It seems like the more time you spend, the better the result, but where do you draw the line?
I’m uninformed but, as a layman, my reaction is, “effort that is equal to how much I want the job” in the context of how my job search has been going so far
Exactly, theres a slippery slope, you cant follow their instructions (eg 8 hours max), do as much as you are willing to do, because thats who you are competing against ,
Take home machine learning model building is a serious waste of time, here’s a week, fuck around and worry if your accuracy is high enough, then get rejected
All they want is XGBoost or a neural network with input categorical encoding or feature generation, then 2 dense layers, output sigmoid, softmax, or linear output. Then they pick the person who has the PHD who just did a Random Forest Gridsearch