Kushal Singh

Key Skills

Software

Telus

Top Subject Matter

No subject matter listed

Top Data Types

Text

Top Label Types

RLHF

Fine Tuning

Evaluation Rating

Prompt Response Writing SFT

Freelancer Overview

I am a mathematics researcher and educator with extensive experience in training and evaluating AI systems, particularly for advanced mathematical reasoning tasks. My background includes over seven years of teaching and research, a Ph.D. in Pure Mathematics, and hands-on work as an AI training and evaluation specialist. I have contributed to projects focused on annotating mathematical reasoning, designing evaluation prompts, and documenting model inconsistencies to enhance AI reliability. I am proficient in tools such as Mathematica, LaTeX, Overleaf, and ChatGPT, and have a strong foundation in mathematical modeling, analysis, and content validation. My passion lies in bridging the gap between mathematics and AI, using my expertise to ensure high-quality, accurate training data for intelligent systems.

Labeling Experience

Prompt–Response Mathematical Reasoning Task

TelusTextRLHFFine Tuning

Contributed to a high-impact project designed to enhance the mathematical reasoning abilities of advanced AI models. Created challenging, multi-skill mathematical prompts and analyzed the model’s responses in detail. Annotated each step of the model’s solution to identify errors in logic, calculation, and notation. Provided corrected solutions and detailed commentary to illustrate conceptual understanding and reasoning accuracy. Delivered high-quality annotated data that supported fine-tuning and evaluation of language models used for mathematics education and automated problem-solving.

Contributed to a high-impact project designed to enhance the mathematical reasoning abilities of advanced AI models. Created challenging, multi-skill mathematical prompts and analyzed the model’s responses in detail. Annotated each step of the model’s solution to identify errors in logic, calculation, and notation. Provided corrected solutions and detailed commentary to illustrate conceptual understanding and reasoning accuracy. Delivered high-quality annotated data that supported fine-tuning and evaluation of language models used for mathematics education and automated problem-solving.

2025 - 2025

Mathematical Reasoning and Text Annotation Project

TelusTextRLHFFine Tuning

Reviewed AI-generated mathematical and scientific responses to ensure factual accuracy, correct notation, and logical consistency. Verified step-by-step reasoning, identified reasoning errors, and labeled data according to detailed rubrics. Created improved prompts and gold-standard answers to guide model retraining and evaluation. Maintained high annotation quality standards with consistency and clarity in documentation, contributing to LLM datasets for educational and problem-solving applications.

Reviewed AI-generated mathematical and scientific responses to ensure factual accuracy, correct notation, and logical consistency. Verified step-by-step reasoning, identified reasoning errors, and labeled data according to detailed rubrics. Created improved prompts and gold-standard answers to guide model retraining and evaluation. Maintained high annotation quality standards with consistency and clarity in documentation, contributing to LLM datasets for educational and problem-solving applications.

2025 - 2025

Education

Work History