Mercor / Scale
Mercor and Scale recruit and pay domain experts to produce AI training data through reinforcement learning from human feedback (RLHF) and evaluations. As a contract writer and reviewer on proprietary research projects for leading frontier AI labs, I provide expertise in marketing, sales, business, and new media.
- Role
- Expert (Writer, Reviewer)
- Focus
- Applied AI · Data
- Years
- 2023–Present
- Audiences
- Enterprise
What data labeling is
Data labeling is human work, and it plays a major role in shaping what AI models actually output. Mercor and Scale (in which Meta holds a 49% stake) help enterprises turn their data into production AI and other revenue-generating applications, such as AI agents and robotics.
As an expert trainer, I have evaluated hundreds of datasets, providing feedback grounded in references and domain expertise on how models can generate better outputs.
What I did
I wrote and edited social and marketing assets, peer-reviewed other trainers’ responses, and gave feedback informed by the lead researcher’s weekly product updates.
What it taught me
This work showed me how RLHF shapes model outputs, and how much value my own expertise and voice bring to the process.