Carmen Fragelli

Key Skills

Software

Doccano

Labelbox

Mindrift

OneForma

Prodigy

Other

Internal/Proprietary Tooling

Top Subject Matter

No subject matter listed

Top Data Types

Audio

Image

Text

Top Task Types

Audio Recording

Prompt Response Writing SFT

Question Answering

Text Generation

Text Summarization

Freelancer Overview

I have experience in data labeling and AI training, gained through my work as an AI Contributor, AI Tutor, and Domain Expert in Education. In these roles, I have evaluated and reviewed AI-generated content, ensuring its ethical and pedagogical accuracy. I have also contributed to the development of reliable and context-aware language models. I have over a year of experience in data labeling and AI training, gained through my work as an AI Contributor, AI Tutor, and Domain Expert in Education. In these roles, I have evaluated and reviewed AI-generated content, ensuring its ethical and pedagogical accuracy. I have also contributed to the development of reliable and context-aware language models. My key skills in this area include data annotation and prompt evaluation for Large Language Models (LLMs) , as well as the design of NLP tasks such as classification, sentiment analysis, and question answering. I have provided feedback to optimize NLP outputs and collaborated on multilingual quality assurance (English-Spanish) to ensure cultural appropriateness. Additionally, I have supported the supervised fine-tuning of LLMs through content review.

Labeling Experience

AI Reasoning and Prompt Complexity

MindriftTextPrompt Response Writing SFT

In this project, my work focused on rewriting and enhancing existing prompts to improve an AI model's reasoning skills. The prompts, based on educational scenarios, were crafted to be complex and nuanced, requiring the AI to go beyond simple information retrieval. The workflow included rewriting dialogues into single, complex prompts, generating and evaluating model failures, creating "golden answers," and defining detailed scoring criteria to ensure the educational merit and accuracy of the final responses.

In this project, my work focused on rewriting and enhancing existing prompts to improve an AI model's reasoning skills. The prompts, based on educational scenarios, were crafted to be complex and nuanced, requiring the AI to go beyond simple information retrieval. The workflow included rewriting dialogues into single, complex prompts, generating and evaluating model failures, creating "golden answers," and defining detailed scoring criteria to ensure the educational merit and accuracy of the final responses.

2025

Evaluation and Refinement of Virtual Educational Assistants

MindriftTextPrompt Response Writing SFT

I worked on an AI evaluation team to review and refine the responses of a virtual assistant designed for students and educators. My role included generating pedagogically relevant prompts and reviewing the content to ensure that the answers were ethical, accurate, and culturally appropriate, aligned with equity in learning principles.

I worked on an AI evaluation team to review and refine the responses of a virtual assistant designed for students and educators. My role included generating pedagogically relevant prompts and reviewing the content to ensure that the answers were ethical, accurate, and culturally appropriate, aligned with equity in learning principles.

2025 - 2025

Multilingual Sentiment Analysis of User Comments

OtherTextData Collection

I participated in a sentiment analysis project to classify user comments from digital platforms in English and Spanish. The main task was to label opinions as positive, negative, or neutral, and to categorize the emotions expressed to help developers optimize interaction on digital culture platforms.

I participated in a sentiment analysis project to classify user comments from digital platforms in English and Spanish. The main task was to label opinions as positive, negative, or neutral, and to categorize the emotions expressed to help developers optimize interaction on digital culture platforms.

2024 - 2024

QA and Fine-Tuning of LLM for Specialized Translation

OtherTextEvaluation Rating

I collaborated on the QA phase of a large language model (LLM) focused on the specialized translation of educational and technical texts from English to Spanish. My role included reviewing the AI-generated translations to ensure accuracy, fluency, and coherence, thus contributing to the supervised fine-tuning of the model.

I collaborated on the QA phase of a large language model (LLM) focused on the specialized translation of educational and technical texts from English to Spanish. My role included reviewing the AI-generated translations to ensure accuracy, fluency, and coherence, thus contributing to the supervised fine-tuning of the model.

2023 - 2024

Education

Work History