Canchola Edgar

Key Skills

Software

Labelbox

Top Subject Matter

No subject matter listed

Top Data Types

Image

Top Label Types

Polygon

Freelancer Overview

I am an analytical and detail-oriented data annotator and AI trainer with extensive experience labeling and categorizing multimodal datasets, including images, video, and audio, to support machine learning model development. My background in mathematics and computer science enables me to excel in complex annotation tasks, such as bounding boxes, semantic segmentation, sentiment analysis, and specialized medical image categorization. I am highly skilled in using annotation tools like Labelbox, CVAT, and Scale AI, and consistently achieve high QA scores by following technical guidelines with precision. I have contributed to refining labeling processes, identifying edge cases, and evaluating AI outputs for factual accuracy and safety, all while meeting demanding project deadlines in remote environments. My commitment to quality and efficiency ensures that the data I prepare drives accurate and reliable AI models.

Labeling Experience

data anotator

LabelboxImagePolygon

Project Overview: The "Street-Smart" Vision InitiativeThe scope of this project was to develop a robust training dataset for an autonomous delivery drone system. We weren't just looking for "cars" and "trees"; we needed to teach the model to understand nuance—the difference between a stationary trash can and a pedestrian standing still, or the distinction between a clear path and a glass door.Specific Data Labeling TasksTo get the level of detail required, our team performed three primary types of annotation:2D Bounding Boxes: Identifying all mobile actors (vehicles, cyclists, pedestrians) to establish spatial awareness.Semantic Segmentation: This was the heavy lifting. We pixel-masked static environments—sidewalks, roads, and "no-go" zones—to ensure the drone understood traversable surfaces.Keypoint Annotation: For human figures, we mapped joints (shoulders, knees, ankles) to help the model predict intent, such as whether a person is about to step off a curb.Project Scale and VolumeTh

Project Overview: The "Street-Smart" Vision InitiativeThe scope of this project was to develop a robust training dataset for an autonomous delivery drone system. We weren't just looking for "cars" and "trees"; we needed to teach the model to understand nuance—the difference between a stationary trash can and a pedestrian standing still, or the distinction between a clear path and a glass door.Specific Data Labeling TasksTo get the level of detail required, our team performed three primary types of annotation:2D Bounding Boxes: Identifying all mobile actors (vehicles, cyclists, pedestrians) to establish spatial awareness.Semantic Segmentation: This was the heavy lifting. We pixel-masked static environments—sidewalks, roads, and "no-go" zones—to ensure the drone understood traversable surfaces.Keypoint Annotation: For human figures, we mapped joints (shoulders, knees, ankles) to help the model predict intent, such as whether a person is about to step off a curb.Project Scale and VolumeTh

2023

Education

Work History