Responsibilities
- Review AI-generated descriptions of model structures, training methods, and gradient computation for technical precision.
- Assess machine learning code and notebooks for correctness, including data handling, training procedures, and evaluation logic.
- Deliver accurate human feedback to improve reinforcement learning from human feedback (RLHF) systems, focusing on safety and usefulness.
- Examine how AI models process multi-step reasoning tasks and pinpoint failures in logical flow.
- Perform side-by-side evaluations of model outputs using defined technical criteria and performance indicators.
Work Arrangement
Remote (Worldwide)
Other
- Tasks may include paid assignments requiring up to one hour of focused effort, though many take less time.
- Work hours are flexible.
- Must be able to operate effectively in a home-based work environment.
.png?1769074621){kind=icon}