The River Cleaner Who Never Sent A CV
Bittu Tabahi did not apply for a job. He cleaned a river, and he kept at it long enough and visibly enough that the work led to a job offer from the Netherlands. No application form, no carefully worded summary at the top of a document, no recruiter in the loop at the start. There was a problem in front of him, a decision to keep showing up for it, and an employer in another country who saw enough in the doing to make an offer to a stranger.
Most people will read that as a feel good story about persistence rewarded. I read it as a hiring story, and a slightly uncomfortable one. An employer looked past every filter our industry spent decades building, the degree, the title history, the neat two page summary, and hired on the only evidence that was never easy to manufacture. The man had visibly done the work. Everything else that normally decides a shortlist turned out to be optional.
The one thing a language model cannot generate on a candidate's behalf is a record of having actually done the thing.
That matters more this year than last, because the cost of producing a convincing application has collapsed to roughly nothing. Every candidate now has access to tools that turn a thin history into fluent prose, which means written application material carries far less information than it did two years ago. When a signal becomes free to produce, it stops being a signal. What has repriced upward is the opposite kind of evidence, work done in the open and unprompted, where the record exists whether or not anyone happened to be hiring.
I see this in the AI and data market every week. The candidates who move fastest are rarely the ones with the most polished profiles. They are the ones who shipped something nobody asked them to ship, and can walk you through the decisions inside it without notes. That is not a personality trait. It is simply the last category of proof that survives contact with a world where anyone can produce a perfect paragraph about themselves in nine seconds.
For hiring leaders, the practical read is simple. Stop treating the application as the primary artifact and start treating it as an index. Ask every serious candidate for one thing they built, fixed or ran that nobody assigned to them, and spend the interview there. It cannot be produced on demand the night before, and it is the closest look you will get at how someone behaves when nobody is grading them.
Planning a senior AI or data hire and want the market view first? We share it either way.
Send us the brief →