For employers

Hire this AI Trainer

Sign in or create an account to invite AI Trainers to your job.

LoginCreate account
Invite to Job
J
Joao Onofrio

Joao Onofrio

Labeled Dataset Creation – FireModel (Geospatial Change Detection)

Brazil flagN/A, Brazil
$50.00/hrIntermediate

Key Skills

Software

No software listed

Top Subject Matter

Geospatial Change Detection
Emotion Recognition/Classification
Time-series Model Preparation

Top Data Types

TextText
DocumentDocument
Computer Code ProgrammingComputer Code Programming

Top Task Types

SegmentationSegmentation
Emotion RecognitionEmotion Recognition

Freelancer Overview

Labeled Dataset Creation – FireModel (Geospatial Change Detection). Brings 2+ years of professional experience across complex professional workflows, research, and quality-focused execution. Core strengths include N and A. Education includes Bachelor of Science, Pontifícia Universidade Católica do Paraná (2024). AI-training focus includes data types such as Geospatial, Tiled Imagery, and Image and labeling workflows including Segmentation, Emotion Recognition, and N.

IntermediateEnglishPortugueseSpanish

Labeling Experience

Dataset Preparation for Time-Series Models (N-BEATS / N-BEATSx)

I prepared and structured time-series datasets for model training with N-BEATS. Tasks included cleaning, normalization, and application of data augmentation to cover edge cases. All work contributed to enhanced model robustness and accurate AI outcomes. • Focused on data quality prior to model training. • Applied pre-processing techniques for reliable time-series inputs. • Ensured input distribution was thoroughly represented. • Supported model improvement with comprehensive dataset preparation.

I prepared and structured time-series datasets for model training with N-BEATS. Tasks included cleaning, normalization, and application of data augmentation to cover edge cases. All work contributed to enhanced model robustness and accurate AI outcomes. • Focused on data quality prior to model training. • Applied pre-processing techniques for reliable time-series inputs. • Ensured input distribution was thoroughly represented. • Supported model improvement with comprehensive dataset preparation.

Not specified

Image Classification Labeling – CK+ Emotion Dataset

ImageEmotion Recognition
I handled labeling tasks for image data used in supervised emotion classification. I validated label consistency and organized the dataset for a NO-LOSO evaluation pipeline. My work spanned both supervised and semi-supervised setups, ensuring clean ground-truth labels for model training. • Emphasized label quality to improve AI model accuracy. • Applied structured annotation practices across the dataset. • Maintained documentation of decisions and handled edge cases. • Supported robust evaluation pipelines for emotion recognition models.

I handled labeling tasks for image data used in supervised emotion classification. I validated label consistency and organized the dataset for a NO-LOSO evaluation pipeline. My work spanned both supervised and semi-supervised setups, ensuring clean ground-truth labels for model training. • Emphasized label quality to improve AI model accuracy. • Applied structured annotation practices across the dataset. • Maintained documentation of decisions and handled edge cases. • Supported robust evaluation pipelines for emotion recognition models.

Not specified

Labeled Dataset Creation – FireModel (Geospatial Change Detection)

Segmentation
I built labeled datasets from Sentinel-2 satellite images for pre/post change detection tasks. Labeled masks were generated by rasterizing vector perimeter data and tiling paired images. I performed quality control to verify dataset balance and foreground ratios before training AI models. • Conducted meticulous annotation following strict guidelines. • Ensured consistency in label application across a large dataset. • Verified data integrity through spot-auditing and dataset checks. • Contributed to domain-specific geospatial AI applications.

I built labeled datasets from Sentinel-2 satellite images for pre/post change detection tasks. Labeled masks were generated by rasterizing vector perimeter data and tiling paired images. I performed quality control to verify dataset balance and foreground ratios before training AI models. • Conducted meticulous annotation following strict guidelines. • Ensured consistency in label application across a large dataset. • Verified data integrity through spot-auditing and dataset checks. • Contributed to domain-specific geospatial AI applications.

Not specified

Education

P

Pontifícia Universidade Católica do Paraná

Bachelor of Science, Computer Science

Bachelor of Science
2024

Work History

C

Capacitant

Tech Lead

N/A
2025 - Present
B

Baldussi telecom

Backend Dev Inter

Curitiba
2024 - 2026