Arslaan Paray

Key Skills

Software

Anno-Mage

Clickworker

Data Annotation Tech

HiveMind

CVAT

Internal/Proprietary Tooling

Top Subject Matter

No subject matter listed

Top Data Types

Computer Code Programming

Document

Text

Top Task Types

Bounding Box

Classification

Computer Programming Coding

Mapping

RLHF

Freelancer Overview

I have hands-on experience with AI training projects using Outlier AI for evaluating AI outputs and multi-modal data annotation, and Alligner for precise 3D object pose labeling supporting autonomous vehicle perception. My core skills include high-performance systems, cloud architecture, infrastructure automation, distributed systems, DevOps, CI/CD, and cybersecurity. I bring strong attention to detail and expertise in optimizing scalable, reliable data pipelines to deliver top-quality annotated data that enhances AI model accuracy.

Labeling Experience

LLM Text Annotation & Quality Assurance with Invisible Technologies

Internal Proprietary ToolingTextEntity Ner ClassificationQuestion Answering

Collaborated with Invisible Technologies on cutting-edge LLM annotation projects, labeling large-scale text datasets for natural language understanding and generation. Conducted quality assurance to ensure annotation accuracy and guideline adherence, improving dataset quality for model fine-tuning and reinforcement learning from human feedback (RLHF). Participated in refining annotation guidelines and evaluating chain-of-thought reasoning to enhance AI model reasoning capabilities.

Collaborated with Invisible Technologies on cutting-edge LLM annotation projects, labeling large-scale text datasets for natural language understanding and generation. Conducted quality assurance to ensure annotation accuracy and guideline adherence, improving dataset quality for model fine-tuning and reinforcement learning from human feedback (RLHF). Participated in refining annotation guidelines and evaluating chain-of-thought reasoning to enhance AI model reasoning capabilities.

2024 - 2025

Indoor Robotics Navigation – 3D Environment Annotation with Alligner

Internal Proprietary Tooling3D SensorPolygonSegmentation

Created high‑precision 3D annotations for indoor robotics navigation systems using RGBD sensor data. Leveraged the Alignerr platform to label objects such as furniture, doors, pathways, and obstacles with cuboids and polygons, and added pose estimation data to improve spatial awareness for autonomous robots. The annotated dataset was used to train SLAM (Simultaneous Localization and Mapping) and object detection models, significantly enhancing navigation accuracy in complex indoor environments. Maintained over 97% annotation accuracy through strict guideline compliance and rigorous quality assurance checks.

Created high‑precision 3D annotations for indoor robotics navigation systems using RGBD sensor data. Leveraged the Alignerr platform to label objects such as furniture, doors, pathways, and obstacles with cuboids and polygons, and added pose estimation data to improve spatial awareness for autonomous robots. The annotated dataset was used to train SLAM (Simultaneous Localization and Mapping) and object detection models, significantly enhancing navigation accuracy in complex indoor environments. Maintained over 97% annotation accuracy through strict guideline compliance and rigorous quality assurance checks.

2024 - 2025

Road Damage Detection Data Annotation Project

CVATImageBounding BoxPolygon

Labeled large sets of street‑level dashcam images to detect and categorize road surface damage, including potholes, cracks, and patches. Used bounding boxes and polygons to accurately mark affected areas and added metadata such as severity level and location. The annotated dataset was used to train a computer vision model for automated road damage detection, aiding municipalities in infrastructure maintenance. Ensured high accuracy by following strict labeling guidelines and performing thorough quality checks throughout the project.

Labeled large sets of street‑level dashcam images to detect and categorize road surface damage, including potholes, cracks, and patches. Used bounding boxes and polygons to accurately mark affected areas and added metadata such as severity level and location. The annotated dataset was used to train a computer vision model for automated road damage detection, aiding municipalities in infrastructure maintenance. Ensured high accuracy by following strict labeling guidelines and performing thorough quality checks throughout the project.

2023 - 2024

Education

Work History