Stefan Petcov

Key Skills

Software

OneForma

Remotasks

Scale AI

Top Subject Matter

No subject matter listed

Top Data Types

Document

Image

Text

Top Task Types

Classification

Evaluation Rating

Prompt Response Writing SFT

Question Answering

Freelancer Overview

Over the past year, I’ve contributed to a range of data labeling and AI training projects through platforms like Remotasks and Outlier AI. My work has included image annotation, text categorization, audio transcription, and quality assurance tasks across multiple domains. This experience has given me a strong foundation in understanding model requirements, following detailed instructions, and maintaining high accuracy and consistency under tight deadlines. I’ve developed a keen eye for detail and a solid grasp of data integrity principles, ensuring that the labeled datasets I work on meet rigorous standards. While I’m early in my journey in the AI space, I’ve quickly adapted to new tools and workflows, and I’m enthusiastic about continuing to grow in this field. I take pride in delivering high-quality work that supports the development of reliable, ethical AI systems.

Labeling Experience

Bulba Languages

Scale AITextTranslation Localization

The Bulba Languages project focused on evaluating and improving the performance of multilingual large language models (LLMs) by assessing AI-generated responses across various languages and cultural contexts. My role involved tasks such as rating the quality, factual accuracy, tone, and cultural appropriateness of model outputs. This required following strict annotation guidelines and quality control procedures to ensure the training data aligned with ethical standards and real-world linguistic nuances. The project was large-scale and collaborative, with contributors across different regions working to fine-tune the model’s behavior. I was responsible for identifying and flagging inappropriate or biased responses, ensuring that the AI system responded accurately and respectfully across different scenarios. Maintaining high consistency and quality was critical, as the outputs directly influenced model performance in sensitive use cases like customer support, education, and public-facin

The Bulba Languages project focused on evaluating and improving the performance of multilingual large language models (LLMs) by assessing AI-generated responses across various languages and cultural contexts. My role involved tasks such as rating the quality, factual accuracy, tone, and cultural appropriateness of model outputs. This required following strict annotation guidelines and quality control procedures to ensure the training data aligned with ethical standards and real-world linguistic nuances. The project was large-scale and collaborative, with contributors across different regions working to fine-tune the model’s behavior. I was responsible for identifying and flagging inappropriate or biased responses, ensuring that the AI system responded accurately and respectfully across different scenarios. Maintaining high consistency and quality was critical, as the outputs directly influenced model performance in sensitive use cases like customer support, education, and public-facin

2024 - 2024

Education

Work History