Parsewave builds coding datasets for AI labs. We need Haskell developers to write hard, real-world coding tasks that current models get wrong. You design the problem, write the solution, and make sure the tests actually catch bad answers. The whole thing is yours.
Tasks need to be hard enough that AI agents fail them. If a senior Haskell dev breezes through it, it’s not what we’re looking for.
Remote, worldwide, no meetings, no minimums. $80/task, $90 for strong ones. ~2 hrs average.
It would be cool to try to automate bumping version ranges in cabals and fixing breakages on hackage.
Every GHC release causes an avalanche of boring commits…