Nesrin Ali

Key Skills

Software

CVAT

Data Annotation Tech

Labelbox

V7 Labs

Internal/Proprietary Tooling

Scale AI

Top Subject Matter

No subject matter listed

Top Data Types

Document

Image

Text

Top Task Types

Classification

Data Collection

Object Detection

Prompt Response Writing SFT

Translation Localization

Freelancer Overview

My experience centers on creating high-fidelity visual data critical for model training, with a specific focus on precision and commercial application. Advanced Segmentation Masking: Consistently deliver pixel-perfect Polygon and Semantic Segmentation Masks for intricate objects and irregular boundaries, maintaining high fidelity required for critical engineering and analytical applications. Domain-Specific Data Creation: Expertise in annotating data for Square Footage (sqft) calculation, demonstrating the ability to interpret complex architectural, structural, and infrastructure imagery (construction, aerial, satellite data). Quality Assurance (QA) Focus: Maintain rigorous quality standards, specifically focusing on strict edge adherence under high magnification to minimize boundary errors and eliminate annotation ambiguity.

Labeling Experience

train datasets for AI models RLHF

Internal Proprietary ToolingDocumentRLHF

training datasets for AI models by applying complex policy guidelines to score and rank LLM-generated responses based on safety, helpfulness, and ethical compliance. Executed comparative evaluations of multiple model outputs to create high-fidelity Conducted rigorous Fact-Checking and Grounding validation, systematically comparing model predictions and generated claims against authoritative, multi-source external data.

training datasets for AI models by applying complex policy guidelines to score and rank LLM-generated responses based on safety, helpfulness, and ethical compliance. Executed comparative evaluations of multiple model outputs to create high-fidelity Conducted rigorous Fact-Checking and Grounding validation, systematically comparing model predictions and generated claims against authoritative, multi-source external data.

2025 - 2025

Cvat image Data Labeling

CVATImageBounding BoxPolygon

used Cvat to extract project details and sqft from thousands of images by labeling the correct sqft.

used Cvat to extract project details and sqft from thousands of images by labeling the correct sqft.

2025 - 2025

Scale AI for LLm

Scale AIDocumentTranslation Localization

Applied complex policy guidelines to evaluate, rank, and score LLM responses based on quality criteria including truthfulness, relevance, completeness, and tone (using concepts like preference modeling). Provided comparative feedback on multiple model generations to create high-quality prompt/response pairs used for fine-tuning the base model.

Applied complex policy guidelines to evaluate, rank, and score LLM responses based on quality criteria including truthfulness, relevance, completeness, and tone (using concepts like preference modeling). Provided comparative feedback on multiple model generations to create high-quality prompt/response pairs used for fine-tuning the base model.

2024 - 2025

Education

Work History