Aliah Saleh

Key Skills

Software

AWS SageMaker

Appen

Labelbox

Surge AI

Scale AI

Top Subject Matter

No subject matter listed

Top Data Types

Document

Image

Text

Top Task Types

Prompt Response Writing SFT

Question Answering

Text Generation

Text Summarization

Translation Localization

Freelancer Overview

As a PhD-level AI Data Labeling and Evaluation Specialist with over seven years of experience, I have led large-scale annotation projects supporting natural language processing (NLP), computer vision, and multilingual AI systems. My expertise covers text, image, audio, and video labeling for major platforms including Labelbox, Scale AI, Appen, and Toloka. I have managed complex projects involving LLM evaluation, sentiment and intent classification, entity recognition, image segmentation, and data quality assurance. My technical proficiency in Python, JSON, and data structuring allows me to bridge the gap between human annotation and automated validation workflows. I bring a research-driven approach from my doctoral work in machine learning, focusing on optimizing human-in-the-loop labeling systems to improve model performance, accuracy, and fairness. 1. LLM Evaluation and Prompt Optimization (English & Arabic) Evaluated and annotated text data for factual accuracy, tone, and coherence in multilingual contexts. Supported reinforcement learning fine-tuning through structured feedback datasets. 2. Computer Vision Annotation for Autonomous Driving Conducted image segmentation and object detection labeling for road scenes. Ensured 99% labeling accuracy through QA validation and cross-review processes. 3. Multilingual Text Classification for NLP Model Training Tagged and categorized text across English, Arabic, and Swahili datasets. Enhanced dataset consistency for improve

Labeling Experience

Image and Video Annotation for Object Detection

Scale AIImageBounding Box

Labeled and annotated large-scale image and video datasets used for autonomous driving and computer vision model training. Tasks included object detection using bounding boxes and polygon segmentation to identify vehicles, pedestrians, road signs, and lane markings. Maintained over 98% accuracy by following strict annotation guidelines and performing multi-layer quality checks. Contributed to datasets exceeding 20,000 frames, enhancing model reliability and environmental awareness.

Labeled and annotated large-scale image and video datasets used for autonomous driving and computer vision model training. Tasks included object detection using bounding boxes and polygon segmentation to identify vehicles, pedestrians, road signs, and lane markings. Maintained over 98% accuracy by following strict annotation guidelines and performing multi-layer quality checks. Contributed to datasets exceeding 20,000 frames, enhancing model reliability and environmental awareness.

2021 - 2023

Education

Work History