How We Work
We're a labor pool for alignment research. Labs and projects come to us to implement evals, run baselines, annotate data, and handle administrative overhead. The idea is simple: researchers focus on research, and we handle the grunt work.
We run on a task-based model. Contributors—a core group of regulars plus a broader pool of specialists—pick up tasks as they're available and work on their own schedule. People work as much or as little as they want, as long as they get their tasks done.
Everyone involved is here because they think alignment matters. That means people actually care whether projects work, not just whether they get done. It's not about hitting milestones or collecting invoices—if something needs fixing or a better approach becomes obvious halfway through, we will say so.
