Fernando Federici

Key Skills

Software

Appen

CVAT

Labelbox

OneForma

SuperAnnotate

Surge AI

Other

Top Subject Matter

Data labeling

Quality Assurance

Annotation Tasks:

Top Data Types

Image

Text

Video

Top Label Types

Bounding Box

Classification

Data Collection

Segmentation

Translation Localization

Freelancer Overview

With solid experience in data labeling, I specialize in annotating and curating text datasets for machine learning and large language models (LLMs). My work involves evaluating model outputs, refining prompts, and applying linguistic expertise to ensure precise annotations. I am proficient in following complex guidelines, performing rigorous quality assurance checks, and contributing to scalable workflows that improve training data reliability.

Labeling Experience

GUMGUM

SuperannotateTextClassificationEvaluation Rating

I review and analyze text content across a wide range of websites to identify and classify threatening or harmful concepts. My role involves carefully reading each article to determine the presence of threats (such as physical violence, terrorism, or self-harm) and classifying the content according to predefined categories and guidelines. When a threat is detected, I verify its specific type and context to ensure consistent labeling and accurate data quality. This project's purpose is to train and refine models that improve the safety and effectiveness of online advertising campaigns, ensuring that ads are not displayed alongside unsafe or harmful content.

I review and analyze text content across a wide range of websites to identify and classify threatening or harmful concepts. My role involves carefully reading each article to determine the presence of threats (such as physical violence, terrorism, or self-harm) and classifying the content according to predefined categories and guidelines. When a threat is detected, I verify its specific type and context to ensure consistent labeling and accurate data quality. This project's purpose is to train and refine models that improve the safety and effectiveness of online advertising campaigns, ensuring that ads are not displayed alongside unsafe or harmful content.

2025

Landmeen

AppenVideoClassificationFine Tuning

In this project, I reviewed and evaluated short video clips to help train and improve an AI model. My main task was to classify videos based on their content and context, following specific guidelines to ensure consistent and high-quality labeling. I assessed visual and contextual elements to determine the appropriate category for each video, contributing to the model’s ability to accurately recognize and understand different types of video content. This work supported the supervised fine-tuning process, ensuring the model learned from precisely labeled examples.

In this project, I reviewed and evaluated short video clips to help train and improve an AI model. My main task was to classify videos based on their content and context, following specific guidelines to ensure consistent and high-quality labeling. I assessed visual and contextual elements to determine the appropriate category for each video, contributing to the model’s ability to accurately recognize and understand different types of video content. This work supported the supervised fine-tuning process, ensuring the model learned from precisely labeled examples.

2025 - 2025

Diamond

OneformaTextClassificationEmotion Recognition

In this project, I labeled text segments with sentiment categories such as Valence, Arousal, and Emotion, following frameworks from cognitive science research on emotions. My work involved reading the target transcription and its surrounding context to accurately assess the emotional tone of the text. I analyzed how positive or negative (valence) and how calm or intense (arousal) the segment was, and then assigned a corresponding discrete emotion label—such as happy, angry, or sad. The annotation was based primarily on the content of the target segment itself, while the context helped ensure a more accurate and nuanced interpretation of sentiment.

In this project, I labeled text segments with sentiment categories such as Valence, Arousal, and Emotion, following frameworks from cognitive science research on emotions. My work involved reading the target transcription and its surrounding context to accurately assess the emotional tone of the text. I analyzed how positive or negative (valence) and how calm or intense (arousal) the segment was, and then assigned a corresponding discrete emotion label—such as happy, angry, or sad. The annotation was based primarily on the content of the target segment itself, while the context helped ensure a more accurate and nuanced interpretation of sentiment.

2025 - 2025

Education

Work History