Shahzad T.

Key Skills

Software

Labelbox

SuperAnnotate

Scale AI

Top Subject Matter

AI Chatbot Response & Fact-Checking (English/Urdu/Sindhi)

RLHF & Prompt Engineering Evaluation (English/Urdu/Sindhi)

Multilingual Data Annotation (English/Urdu/Sindhi)

Top Data Types

Text

Image

Document

Top Task Types

Evaluation/Rating

Red Teaming

RLHF

Question Answering

Freelancer Overview

I am a dedicated AI Content Evaluator with native fluency in English, Urdu, and Sindhi. I specialize in Reinforcement Learning from Human Feedback (RLHF), fact-checking, and linguistic validation. My focus is on providing high-quality, nuanced data labeling that ensures AI models are culturally accurate and grammatically precise across all three languages. Though I come from a generalist background, I pride myself on my high attention to detail and my ability to handle complex categorization and prompt evaluation tasks. I am highly responsive to following project guidelines strictly, and am committed to helping build safer and more helpful AI systems through consistent, high-accuracy data annotation.

Labeling Experience

Bilingual Translation & Contextualization | Freelance / Personal Projects

Text

Translated and contextualized over sixty texts among English, Sindhi, and Urdu for accuracy and nuance. Paid close attention to cultural and linguistic details during translation. Ensured original intent and meaning were preserved throughout all annotated texts. • Developed expertise in handling text-heavy documents requiring nuanced comprehension. • Demonstrated data entry accuracy and systematic processing of translated materials. • Enhanced skills in detailed review and approval of annotated translations. • Contributed to improved training data quality through precise localization efforts.

Present

AI Interaction & Prompt Evaluation | Independent Study

Text

Independently conducted bilingual stress tests and evaluations on AI platforms to assess model capabilities in Sindhi and Urdu. Reviewed and critiqued numerous AI-generated responses, focusing on logical structure, hallucination tendencies, and cultural context. Applied rigorous standards for evaluating AI fact-checking, validation, and adherence to instructions. • Utilized skills in red-teaming to challenge language model safety and consistency. • Developed evaluation protocols based on RLHF principles and best practices. • Focused on identifying and describing hallucination patterns and cultural inaccuracies. • Maintained detailed records of results and insights to inform AI model improvements.

Present

Education

Work History