Esther Samuel

Key Skills

Software

Appen

Data Annotation Tech

OneForma

Top Subject Matter

No subject matter listed

Top Data Types

Image

Text

Video

Top Task Types

Evaluation/Rating

Prompt + Response Writing (SFT)

Question Answering

Text Generation

Freelancer Overview

Over the past two years, I’ve worked hands-on in data annotation and AI training, contributing to projects that require accuracy, consistency, and strong attention to detail. My experience includes response correctness evaluation, large language model, pairwise caption comparison, and labeling data for self-driving video. These projects has helped train models to better understand language, visual content, and real-world driving environments. I’ve become comfortable working with detailed guidelines, identifying subtle errors, and making judgment calls that directly improve model performance. What sets me apart is my ability to combine careful analytical thinking with clear communication and reliable execution. I enjoy diving into complex tasks, learning quickly, and refining my approach based on feedback. I take pride in producing clean, high-quality annotations and contributing to datasets that make AI systems safer and more effective.

Labeling Experience

Standalone visual appeal with error tagging

Data Annotation TechImageQuestion AnsweringObject Detection

In the Standalone Visual Appeal with Error Tagging project, my main responsibility was to review images and assess their overall visual quality while also identifying any errors or inconsistencies. This included checking for issues like poor composition, blurriness, incorrect cropping, missing elements, or visual artifacts that could affect how the model interpreted the image. Each image had to be evaluated carefully not just for aesthetic appeal but also for technical accuracy, so I spent a lot of time looking closely at details and applying the project’s guidelines consistently. The project involved a fairly large dataset, so staying organized and maintaining a steady pace without sacrificing quality was extremely important. To ensure accuracy, we followed strict quality measures such as double-checking edge cases, adhering to annotation rules, and meeting required precision benchmarks in our reviews.

In the Standalone Visual Appeal with Error Tagging project, my main responsibility was to review images and assess their overall visual quality while also identifying any errors or inconsistencies. This included checking for issues like poor composition, blurriness, incorrect cropping, missing elements, or visual artifacts that could affect how the model interpreted the image. Each image had to be evaluated carefully not just for aesthetic appeal but also for technical accuracy, so I spent a lot of time looking closely at details and applying the project’s guidelines consistently. The project involved a fairly large dataset, so staying organized and maintaining a steady pace without sacrificing quality was extremely important. To ensure accuracy, we followed strict quality measures such as double-checking edge cases, adhering to annotation rules, and meeting required precision benchmarks in our reviews.

2024 - 2025

Education

Work History