Ansh Raj

Key Skills

Software

OneForma

Other

Data Annotation Tech

Top Subject Matter

No subject matter listed

Top Data Types

Text

Image

Top Label Types

Evaluation Rating

RLHF

Prompt Response Writing SFT

Freelancer Overview

I have practical experience working on AI training data and data labeling tasks, mainly focused on text-based projects such as classification, response evaluation, ranking, and instruction-following assessments. I have completed trial and starter assessments on platforms like DataAnnotation.tech and REX, where I worked on understanding prompts carefully, comparing multiple responses, and applying detailed guidelines to choose or improve the best possible output. These tasks required strong attention to detail, clear reasoning, and consistency to meet quality standards. What sets me apart is my strong language ability and thoughtful approach to evaluation. With a background in teaching and writing, I am comfortable working with nuanced text, identifying errors, and explaining decisions clearly. I adapt quickly to new project rules, manage tasks independently, and maintain accuracy even under time constraints. I take quality seriously and aim to contribute data that genuinely helps improve AI model performance, rather than just completing tasks mechanically.

Labeling Experience

ranking and comparison tasks

Data Annotation TechTextPrompt Response Writing SFT

I have participated in ranking and comparison tasks, where responses were scored or ranked against each other using predefined criteria, and quality review/error analysis projects, identifying factual errors, logical gaps, or guideline violations and providing corrective feedback. Across these projects, my focus has been on producing high-quality, consistent annotations that directly improve model performance, particularly in language understanding, reasoning, and instruction adherence.

I have participated in ranking and comparison tasks, where responses were scored or ranked against each other using predefined criteria, and quality review/error analysis projects, identifying factual errors, logical gaps, or guideline violations and providing corrective feedback. Across these projects, my focus has been on producing high-quality, consistent annotations that directly improve model performance, particularly in language understanding, reasoning, and instruction adherence.

2025

instruction-following and response evaluation projects

OtherImageRLHFEvaluation Rating

I have worked on a range of text-focused AI training and data labeling projects that required careful judgment, consistency, and adherence to detailed guidelines. These include instruction-following and response evaluation projects, where I reviewed multiple AI-generated answers to a single prompt and selected or refined the best response based on accuracy, relevance, tone, and safety. I have also handled text classification and categorization tasks, labeling content by intent, topic, sentiment, or quality level to support supervised learning for NLP models.

I have worked on a range of text-focused AI training and data labeling projects that required careful judgment, consistency, and adherence to detailed guidelines. These include instruction-following and response evaluation projects, where I reviewed multiple AI-generated answers to a single prompt and selected or refined the best response based on accuracy, relevance, tone, and safety. I have also handled text classification and categorization tasks, labeling content by intent, topic, sentiment, or quality level to support supervised learning for NLP models.

2025

AI Trainer / AI Content Evaluator

OneformaTextEvaluation Rating

Reviewed and evaluated AI-generated text for compliance and accuracy. Corrected outputs to ensure adherence to instructions. Evaluated content for bias and safety. Adapted to new evaluation tools quickly. • Identified and removed hallucinations and factual errors. • Compared multiple AI responses to select the highest-quality output. • Handled ambiguous prompts with consistent logic. • Maintained high-quality scores (95%+ target) through careful review.

Reviewed and evaluated AI-generated text for compliance and accuracy. Corrected outputs to ensure adherence to instructions. Evaluated content for bias and safety. Adapted to new evaluation tools quickly. • Identified and removed hallucinations and factual errors. • Compared multiple AI responses to select the highest-quality output. • Handled ambiguous prompts with consistent logic. • Maintained high-quality scores (95%+ target) through careful review.

2023 - 2025

Education

Work History