Data Annotation Tech - Rubric Writing
Data Annotation TechTextTranslation LocalizationRLHF
In this project, I designed complex prompts to intentionally trigger model weaknesses and evaluate its behavior under challenging scenarios. By applying prompt engineering and red teaming techniques, I identified edge cases and inconsistencies in the model’s reasoning. I then developed detailed rubrics to assess response quality and applied RLHF principles to guide the model toward safer, more accurate, and instruction-aligned outputs. This work strengthened my ability to analyze model performance, define evaluation standards, and improve large language model reliability.
In this project, I designed complex prompts to intentionally trigger model weaknesses and evaluate its behavior under challenging scenarios. By applying prompt engineering and red teaming techniques, I identified edge cases and inconsistencies in the model’s reasoning. I then developed detailed rubrics to assess response quality and applied RLHF principles to guide the model toward safer, more accurate, and instruction-aligned outputs. This work strengthened my ability to analyze model performance, define evaluation standards, and improve large language model reliability.
2025