[mlthoughts2018]

I don’t know about Salesforce specifically, but this type of thing is surprisingly common.

If your first interaction with candidates, even from a huge, huge open pipeline of applications, is an automatic request for a time-intensive project or skills assessment, something is very broken with that hiring process.

You simply have to pay the (high & difficult) cost of evaluating & filtering applications and then having a detailed conversation with the small number of applicants you choose to move forward with. You need to understand their technical interests and match to the role, and let them ask lots of questions to even know if they are interested enough to invest time in the rest of the interviewing.

Personally one major change I’ve instituted over a few years in the ML & data science pipeline at my company is that we only do a technical conversational interview, no shared screen coding or take home project at all, it just doesn’t tell you anything, and counterintuitively you end up wasting more time and money administering it than if you just bring people onsite sooner and bite the bullet to evaluate them in a full interview day.