Cristian Cossia

Key Skills

Software

Data Annotation Tech

OneForma

Remotasks

Scale AI

Toloka

Telus

Top Subject Matter

No subject matter listed

Top Data Types

3D Sensor

Document

Image

Top Task Types

Bounding Box

Classification

Object Detection

Polygon

Segmentation

Freelancer Overview

I have solid experience in data labeling and AI training, having worked on well-known platforms such as Remotasks, OneForma, Scale AI, and others. I've contributed to various projects performing tasks such as IQA, image segmentation, polygon annotation, and visual content classification. I've also worked on projects involving LLMs in Spanish, which included tasks such as prompt evaluation, response ranking, and quality control for natural language understanding and generation. My focus has always been on maintaining high quality standards, adhering closely to guidelines, and adapting quickly to new annotation schemes. What sets me apart is my attention to detail, my ability to spot errors or inconsistencies in data, and my experience working remotely with teams under tight deadlines. With my technical background and methodical approach, I've helped improve the quality of datasets used in both computer vision and large scale language models, ensuring accurate and reliable data for optimal AI model performance.

Labeling Experience

LLM Spanish

Data Annotation TechTextQuestion AnsweringText Generation

The project focused on LLMs in Spanish, where I was responsible for evaluating prompts and responses. My tasks included reviewing the quality of the model's output, giving ratings based on clarity, relevance, and correctness, and suggesting edits to improve the overall response. This helped enhance the model's natural language understanding and generation capabilities in Spanish.

The project focused on LLMs in Spanish, where I was responsible for evaluating prompts and responses. My tasks included reviewing the quality of the model's output, giving ratings based on clarity, relevance, and correctness, and suggesting edits to improve the overall response. This helped enhance the model's natural language understanding and generation capabilities in Spanish.

2024

IQA

Scale AIImageBounding Box

In the project i was labeling textbook style PDF pages using bounding boxes to identify questions, images, and answers. The task involved two formats: IQA and QQQ_aaa. I followed a linking workflow to ensure accurate connections, supporting structured data for AI model training in math problem analysis.

In the project i was labeling textbook style PDF pages using bounding boxes to identify questions, images, and answers. The task involved two formats: IQA and QQQ_aaa. I followed a linking workflow to ensure accurate connections, supporting structured data for AI model training in math problem analysis.

2024 - 2024

Image Segmentation

Don T DiscloseImageBounding BoxSegmentation

The project focused on image segmentation, where I was responsible for accurately outlining human figures in photos using segmentation tools. The goal was to precisely identify and label people within various scenes to support the development of computer vision models for human detection and recognition.

The project focused on image segmentation, where I was responsible for accurately outlining human figures in photos using segmentation tools. The goal was to precisely identify and label people within various scenes to support the development of computer vision models for human detection and recognition.

2023 - 2024

Polygon

Scale AIImagePolygonSegmentation

The project involved detailed polygon annotation for image segmentation tasks. I used precise polygon tools to outline objects and regions of interest within images, ensuring high accuracy and adherence to labeling guidelines. This data supported computer vision models in tasks like object detection and scene understanding.

The project involved detailed polygon annotation for image segmentation tasks. I used precise polygon tools to outline objects and regions of interest within images, ensuring high accuracy and adherence to labeling guidelines. This data supported computer vision models in tasks like object detection and scene understanding.

2022 - 2022

Education

Work History