Responsibilities
- Review AI-generated descriptions of model architectures, loss functions, and backpropagation to ensure technical correctness
- Verify machine learning code and Jupyter notebooks for accuracy and optimization, including training pipelines and data handling scripts
- Deliver precise human feedback to refine reinforcement learning from human feedback (RLHF) systems and improve model alignment
- Examine how AI models process multi-step reasoning prompts and identify logical inconsistencies or failures
- Perform side-by-side evaluation of model outputs using defined technical criteria and performance benchmarks
Work Arrangement
Remote (Worldwide) — Mesa, US
Other
- Must be available to complete paid tasks requiring up to one hour of focused, uninterrupted work, though most assignments are shorter
- Flexible scheduling allowed
- Work from home
.png?1769074621){kind=icon}