The Cost of Getting a Job Today | Weli's blog
Short summary
A hiring manager argues that requiring job candidates to use paid AI tools during technical interviews creates an unfair burden on people without income, and suggests practical ways to reduce that financial gatekeeping without compromising the hiring process.
- Requiring candidates to use paid LLM tools for coding challenges is a new financial barrier unique to hiring—unlike housing or food, these costs exist only to get the job and don't benefit the candidate otherwise.
- Moving technical challenges later in the interview process and making them smaller/free-tier compatible reduces the cost burden on candidates who are already financially vulnerable.
- Good engineers are lost to hiring processes that screen out people who can't afford subscription AI tools, even when they'd be valuable team members—a cost companies never see.
Full summary
A hiring manager at a startup noticed something uncomfortable: requiring job candidates to use AI tools during technical interviews creates an unexpected financial barrier. When someone's unemployed and interviewing with multiple companies simultaneously, they're potentially spending money on LLM subscriptions or API credits just to complete coding challenges—at precisely the moment they can least afford it. The author points out this cost is different from other job-search expenses like transportation or clothing. Those things have value beyond hiring. LLM access exists solely to pass a company's screening process, making it a pure tax on desperation.
The author wrestled with potential fixes and found most don't work. Offering candidates a company account for the challenge creates liability issues. Reimbursing their spending gets messy fast—it's hard to justify internally and quickly becomes a gray area. So instead of trying to eliminate the cost entirely, they focused on what they could actually control: the hiring process itself.
Two practical changes: delay the technical challenge until after a real conversation has happened, so candidates only spend money on companies genuinely interested in them. Second, make the challenge small enough to work on free tiers or basic plans, then spend the actual interview time discussing their approach. The author argues the conversation reveals far more than watching someone burn through tokens on an expensive model. The underlying question for any hiring team: how much is your process costing someone with zero income right now, and how many good engineers are you losing without knowing it?