Cosmo
ImageRLHF
Here’s a clean way to describe it in your style:
---
I worked on a project where I did RLHF using prompts, mainly focused on improving AI responses. My role was to create high-quality prompts, review model outputs, and guide the model towards better reasoning, accuracy, and clarity.
I compared different responses, ranked them, and gave detailed feedback on which one was better and why. I also rewrote weak answers to make them more correct, consistent, and aligned with guidelines.
The work required strong attention to detail, critical thinking, and the ability to spot subtle errors in reasoning, not just grammar. Overall, I helped improve how the model understands prompts and produces reliable, high-quality answers.
Here’s a clean way to describe it in your style: --- I worked on a project where I did RLHF using prompts, mainly focused on improving AI responses. My role was to create high-quality prompts, review model outputs, and guide the model towards better reasoning, accuracy, and clarity. I compared different responses, ranked them, and gave detailed feedback on which one was better and why. I also rewrote weak answers to make them more correct, consistent, and aligned with guidelines. The work required strong attention to detail, critical thinking, and the ability to spot subtle errors in reasoning, not just grammar. Overall, I helped improve how the model understands prompts and produces reliable, high-quality answers.
2026 - Present