Thiago Laguna

Key Skills

Software

Labelbox

Top Subject Matter

No subject matter listed

Top Data Types

Image

Text

Video

Top Task Types

Action Recognition

Audio Recording

Bounding Box

Classification

Data Collection

Freelancer Overview

Computer vision specialist with 2 years of experience in data labeling, focused on AI driven imagery and video analysis using Labelbox. From 2023-2025, I've annotated diverse datasets including overhead and geospatial visuals, emphasizing object detection, segmentation, and classification to enhance models for urban and environmental applications. A few generalist image/video and audio jobs too.

Labeling Experience

Video generation labelling and tasking

LabelboxVideoFine TuningEvaluation Rating

In this ongoing project (started November 2025), I evaluate and rate user-submitted videos of everyday household tasks, such as cleaning, cooking, and organizing, captured in home environments. Using Labelbox, I annotate key elements like objects, actions, and sequences to train AI models for smart home automation.

In this ongoing project (started November 2025), I evaluate and rate user-submitted videos of everyday household tasks, such as cleaning, cooking, and organizing, captured in home environments. Using Labelbox, I annotate key elements like objects, actions, and sequences to train AI models for smart home automation.

2025

Sea Satelite Imagery

LabelboxImageBounding BoxClassification

annotated satellite imagery datasets to detect and classify vessels in oceanic environments, processing over 1,000 high-resolution images via Labelbox. Tasks included drawing bounding boxes around ships and categorizing them by type (e.g., cargo, fishing) to train AI models for maritime monitoring. Quality was maintained through double-review protocols and consistency checks, achieving 96% annotation accuracy against benchmarks.

annotated satellite imagery datasets to detect and classify vessels in oceanic environments, processing over 1,000 high-resolution images via Labelbox. Tasks included drawing bounding boxes around ships and categorizing them by type (e.g., cargo, fishing) to train AI models for maritime monitoring. Quality was maintained through double-review protocols and consistency checks, achieving 96% annotation accuracy against benchmarks.

2025 - 2025

Prompt fine tuning

LabelboxTextText GenerationFine Tuning

I fine-tuned prompts for LLM training by evaluating and rating text-based datasets

I fine-tuned prompts for LLM training by evaluating and rating text-based datasets

2024 - 2025

Education

Work History