Work on the data problem, not around it
Small team, building the pipeline that makes AI training data auditable: consented collection, measured annotation, and QC that catches mistakes while they're still cheap to fix.
Open roles
Everything we're hiring for right now. Write in even if nothing below fits.
How we work
Four things that show up in how decisions actually get made here.
Ship the pilot before the scale
We test a new hire on a small, real piece of work before either side commits further. The same standard we hold clients to.
Say the number
If you don't know the accuracy, the timeline or the cost, say so. A guess dressed up as a fact costs more later, for a client and for a teammate.
Own the outcome, not the ticket
Tickets close. A dataset either helps someone's model or it doesn't. That's what people here are judged against.
Written down beats remembered
Guidelines, rubrics and decisions live in a document, not one person's head. It's the only way quality survives someone going on leave.
How hiring works
Five steps, most of them faster than they sound.
- 01
Write in
Tell us what you'd want to work on and what you've actually shipped. No cover letter template needed.
- 02
A short call
Thirty minutes, with whoever you'd actually work with day to day.
- 03
A paid work sample
A small, real piece of work, scoped like a client pilot. Paid regardless of outcome.
- 04
Team conversation
Meet two or three more people you'd be working with.
- 05
A decision
A yes or a no within a week of the work sample, with the reason either way.
Looking for task-based work instead?
The roles above are salaried or contract, full-time. If you want to get paid per task in your own language instead, that’s a different door.
Don't see the right role above?
Write in anyway with what you'd want to work on. We read everything ourselves.