Bashar

Twaij

Bilingual Generative AI Specialist & RLHF Data Evaluator I specialize in evaluating, refining, and optimizing Generative AI outputs with a focus on both Standard Arabic and regional nuances, alongside professional English communication. With a background that bridges practical AI prompting, content analysis, and systematic workflows, I help AI labs and businesses improve model performance, safety, and alignment. Core Expertise: • RLHF & Model Evaluation: Conducting side-by-side evaluations to score responses on accuracy, tone, safety, and instruction-following. • Prompt Engineering & Testing: Designing clear, structured prompts to test model limits, edge cases, and conversational flows. • Fact-Checking & Quality Assurance: Detecting hallucinations, verifying factual accuracy, and eliminating biased or low-quality data. • Localization & Cultural Nuance: Ensuring Arabic outputs align naturally with native expressions, cultural context, and specialized terminology.