Everyone's hiring FDEs.
Nobody can tell who's good.
Every AI lab is copying the Palantir playbook and paying $300k–$1.2m for engineers who can walk into a customer's broken environment and make things work. Job postings grew 800% in nine months.
There's a leaderboard for algorithms. There isn't one for the actual job. So we're building it.
- 01Build. A messy dataset, a vague ask, a deadline. Top scores qualify.
- 02Defend. Explain it to a stakeholder who doesn't care about your architecture. This round decides your rank.
- 03AI allowed. You'll use it on the job, so we score how well you use it. Nothing to hide.
Reserve your handle
Free. First come, first served. Reserved handles get first access to challenge one.
Done. @you is yours.
You're in before the ladder exists. We'll email you when the first challenge is ready.
Until then: the fastest way to move up the waitlist is to bring the people you'd least like to compete against.
Post it on XEmpty until season zero runs. One of these rows could have your name on it, which is the whole point.
Why the second round decides your rank
Here's what everyone who's done this job knows: the code was never the hard part. MIT studied 300 enterprise AI deployments last year and found 95% produced no measurable impact. The models worked. The deployments didn't.
The gap is a person who can find the real blocker — the one the customer can't articulate — build a working slice through the mess, and then sell it to a room that's been burned before. That skill has no leaderboard, no benchmark, no signal beyond "trust me, I was at Palantir."
So the technical round gets you in the door, and defending your build is what you're ranked on. Scored by judges and by the people you're competing against. In public.