Khuram Aziz

Key Skills

Software

Appen

CrowdSource

Data Annotation Tech

Labelbox

Mindrift

OneForma

Telus

Other

Internal/Proprietary Tooling

Top Subject Matter

No subject matter listed

Top Data Types

Audio

Text

Video

Top Task Types

Evaluation Rating

Fine Tuning

Prompt Response Writing SFT

RLHF

Translation Localization

Freelancer Overview

I have over three years of hands-on experience working in AI training data, specifically in data labeling and prompt engineering for large language model development as well as rubric evaluation and audio annotation. My work has focused on Reinforcement Learning from Human Feedback (RLHF) and Supervised Fine Tuning (SFT), where I provided high-quality, human-verified data to refine AI behavior and output. I’ve collaborated on projects for major tech clients through platforms like Invisible, Outlier and Alignerr, among others, where I've worked with large datasets across natural language processing (NLP), sentiment analysis and contextual ranking. What sets me apart is my hybrid background in journalism and digital content, which gives me a deep understanding of tone, nuance, and factual accuracy—key qualities when training AI models to communicate effectively. I bring sharp attention to detail, strong critical thinking, and the ability to balance speed with precision, all while adhering to strict annotation and data quality guidelines. My experience also includes content evaluation, prompt development and multilingual QA support (English, French, Urdu), allowing me to contribute across global projects with cultural sensitivity and linguistic accuracy.

Labeling Experience

TELUS

TelusTextText GenerationRLHF

Worked on prompt and response writing, rubric-based evaluation, and fine-tuning tasks to improve large language model performance. Conducted rigorous QA checks to ensure data quality, consistency, and alignment with project guidelines.

Worked on prompt and response writing, rubric-based evaluation, and fine-tuning tasks to improve large language model performance. Conducted rigorous QA checks to ensure data quality, consistency, and alignment with project guidelines.

2025 - 2025

Innodata

Internal Proprietary ToolingTextClassificationQuestion Answering

Labeled and ranked AI-generated responses based on emotional tone, accuracy, and appropriateness. Contributed to fine-tuning models by evaluating sentiment detection, writing prompts, and providing quality feedback to improve emotional intelligence in AI systems.

Labeled and ranked AI-generated responses based on emotional tone, accuracy, and appropriateness. Contributed to fine-tuning models by evaluating sentiment detection, writing prompts, and providing quality feedback to improve emotional intelligence in AI systems.

2025 - 2025

Alignerr

LabelboxAudioClassificationEmotion Recognition

Annotated and labeled audio clips for precise utterance placement to improve speech recognition accuracy. Ensured high-quality data by performing detailed segmentation, timestamp alignment, and quality checks to support AI model training.

Annotated and labeled audio clips for precise utterance placement to improve speech recognition accuracy. Ensured high-quality data by performing detailed segmentation, timestamp alignment, and quality checks to support AI model training.

2025 - 2025

Invisible

OtherTextClassificationQuestion Answering

Training Microsoft AI and xAI LLM models using a combination of SFT and RLHF.

Training Microsoft AI and xAI LLM models using a combination of SFT and RLHF.

2023 - 2024

Education

Work History