Zainab Anjum

Key Skills

Software

Clickworker

CVAT

Labelbox

OneForma

Remotasks

Scale AI

SuperAnnotate

Toloka

Telus

Other

Top Subject Matter

General

Top Data Types

Document

Image

Text

Top Label Types

Evaluation Rating

Fine Tuning

Prompt Response Writing SFT

RLHF

Text Generation

Freelancer Overview

I have hands-on experience in data labeling and AI training, having worked on image annotation projects and test-phase sample rating tasks for machine learning datasets. My work involved identifying and tagging visual elements with precision, following detailed guidelines to ensure high-quality labeled data for supervised learning models. In addition to annotation, I bring a background in content creation and editorial research, which has strengthened my attention to detail, pattern recognition, and ability to follow complex instructions, all essential in AI training workflows. This combination of analytical thinking and language proficiency makes me a strong fit for diverse AI data roles.

Labeling Experience

Document Annotation Specialist

LabelboxDocumentBounding BoxSegmentation

Annotated complex documents by creating precise header, body, and footer layout zones, ensuring complete page coverage without overlap. Labeled paragraphs, tables, figures, and equations according to strict structural, OCR, and reading-order guidelines to support high-quality document understanding datasets.

Annotated complex documents by creating precise header, body, and footer layout zones, ensuring complete page coverage without overlap. Labeled paragraphs, tables, figures, and equations according to strict structural, OCR, and reading-order guidelines to support high-quality document understanding datasets.

2025

Command prompter

OtherVideoData Collection

Capture long-form (15–30 min) software operation recordings using OBS, completing tasks across command-line tools, CAD environments, editing software, and game development platforms for Taggr ingestion. Perform precise, step-by-step software workflows and maintain recording quality, visibility, and metadata accuracy for downstream training tasks.

Capture long-form (15–30 min) software operation recordings using OBS, completing tasks across command-line tools, CAD environments, editing software, and game development platforms for Taggr ingestion. Perform precise, step-by-step software workflows and maintain recording quality, visibility, and metadata accuracy for downstream training tasks.

2025

Prompt Writer

OtherVideoQuestion AnsweringPrompt Response Writing SFT

Engineer audio-visual reasoning prompts for long-video datasets, forcing LLMs to combine speech, motion, temporal cues, and world knowledge, optimizing for model failure discovery. Conducted adversarial evaluation by comparing ground-truth answers with model predictions to detect hallucinations, missed cues, temporal errors, and multimodal inconsistencies. Classified Q&A samples using multi-label ontology categories and added precise timestamps to strengthen downstream multimodal training data coverage.

Engineer audio-visual reasoning prompts for long-video datasets, forcing LLMs to combine speech, motion, temporal cues, and world knowledge, optimizing for model failure discovery. Conducted adversarial evaluation by comparing ground-truth answers with model predictions to detect hallucinations, missed cues, temporal errors, and multimodal inconsistencies. Classified Q&A samples using multi-label ontology categories and added precise timestamps to strengthen downstream multimodal training data coverage.

2025

LLM Trainer (Videogaming/egocentric/robotic annotation)

OtherVideoAction RecognitionEvaluation Rating

Worked on large-scale robotics and egocentric video understanding campaigns, producing 1,052–1,500+ precision-based annotations across object interaction, spatial reasoning, and multi-step task sequences while maintaining 97% accuracy and adherence to detailed guideline frameworks. Used Grammarly, Gemini, and ChatGPT to normalize phrasing, reduce ambiguity, and streamline batch QA passes, improving annotation speed and cross-file consistency. Performed in-game and first-person perspective (FPP) action analysis for video game playthrough datasets, capturing player intent, environmental context, and action progression to support downstream computer vision and behavioral modeling research.

Worked on large-scale robotics and egocentric video understanding campaigns, producing 1,052–1,500+ precision-based annotations across object interaction, spatial reasoning, and multi-step task sequences while maintaining 97% accuracy and adherence to detailed guideline frameworks. Used Grammarly, Gemini, and ChatGPT to normalize phrasing, reduce ambiguity, and streamline batch QA passes, improving annotation speed and cross-file consistency. Performed in-game and first-person perspective (FPP) action analysis for video game playthrough datasets, capturing player intent, environmental context, and action progression to support downstream computer vision and behavioral modeling research.

2025

Prompt Engineer

OtherTextRLHFPrompt Response Writing SFT

Assessed and scored 50 model responses per week for overall quality and system-instruction adherence, providing structured feedback to improve clarity, factual correctness, and response formatting. Maintained consistent prompt approval and on-time weekly task completion by applying a 5-dimension rubric (relevance, specificity, detail, formatting, and presentation) and iteratively revising prompts based on reviewer feedback.

Assessed and scored 50 model responses per week for overall quality and system-instruction adherence, providing structured feedback to improve clarity, factual correctness, and response formatting. Maintained consistent prompt approval and on-time weekly task completion by applying a 5-dimension rubric (relevance, specificity, detail, formatting, and presentation) and iteratively revising prompts based on reviewer feedback.

2024

Show more

Education

Work History