For employers

Hire this AI Trainer

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

LoginCreate account
Invite to Job
Muhammad Yusuf Fadhe Marwiji

Muhammad Yusuf Fadhe Marwiji

GIS Technical Expert - Geospatial AI

INDONESIA flag
Makssar, Indonesia
$15.00/hrIntermediateAws SagemakerOther

Key Skills

Software

AWS SageMakerAWS SageMaker
Other

Top Subject Matter

No subject matter listed

Top Data Types

Computer Code ProgrammingComputer Code Programming
DocumentDocument
Geospatial Tiled ImageryGeospatial Tiled Imagery
ImageImage
TextText

Top Label Types

Classification
Computer Programming Coding
Data Collection
Geocoding
Land Cover Classification
Mapping
Object Detection
Point Key Point
Polygon
Polyline
Prompt Response Writing SFT

Freelancer Overview

As a Geospatial AI (GeoAI) Specialist and Lecturer with a Master’s degree (GPA 4.00/4.00), I specialize in high-precision data annotation and validation for environmental and earth science models. My expertise lies in creating and refining "ground truth" datasets for supervised machine learning, specifically in Land Use/Land Cover (LULC) classification, object detection in aerial imagery, and disaster risk modeling. Unlike generalist annotators, I bring deep domain knowledge in forestry, hydrology, and remote sensing. I am proficient in using GIS tools (ArcGIS, QGIS, GEE) to generate training data (polygons/points) and evaluating AI model outputs for scientific accuracy, spatial logic, and contextual relevance. I have successfully led projects improving analytical efficiency by 20% through rigorous data quality assurance.

IntermediateEnglishIndonesian

Labeling Experience

Multi-Class Semantic Segmentation for Land Cover Analysis (LULC)

OtherGeospatial Tiled ImageryClassificationMapping
Executed large-scale Semantic Segmentation and pixel-level classification of satellite imagery (Sentinel-2 & Landsat) to generate training data for agricultural planning models. Key AI/ML Workflows: Annotation & Segmentation: Created high-precision polygon masks (Regions of Interest/ROI) for 5+ distinct land cover classes (e.g., Paddy Fields, Primary Forest, Water Bodies, Built-up Areas). Model Training: Utilized Google Earth Engine (GEE) to train Supervised Learning classifiers (Random Forest and CART) based on the annotated ROIs. Accuracy Assessment: Evaluated model performance using Confusion Matrix analysis, achieving an Overall Accuracy of 92% and a Kappa Coefficient of 0.89. Data Cleaning: Performed morphological operations to remove "salt-and-pepper" noise from the classification output, ensuring clean vector data for downstream analysis.

Executed large-scale Semantic Segmentation and pixel-level classification of satellite imagery (Sentinel-2 & Landsat) to generate training data for agricultural planning models. Key AI/ML Workflows: Annotation & Segmentation: Created high-precision polygon masks (Regions of Interest/ROI) for 5+ distinct land cover classes (e.g., Paddy Fields, Primary Forest, Water Bodies, Built-up Areas). Model Training: Utilized Google Earth Engine (GEE) to train Supervised Learning classifiers (Random Forest and CART) based on the annotated ROIs. Accuracy Assessment: Evaluated model performance using Confusion Matrix analysis, achieving an Overall Accuracy of 92% and a Kappa Coefficient of 0.89. Data Cleaning: Performed morphological operations to remove "salt-and-pepper" noise from the classification output, ensuring clean vector data for downstream analysis.

2025 - 2025

Supervised ML Training Data for Flood Susceptibility Prediction

OtherGeospatial Tiled ImageryClassificationMapping
Managed the end-to-end creation of a labeled geospatial dataset used to train and validate Supervised Machine Learning models (Random Forest, SVM, and XGBoost) for flood risk prediction. Key AI/ML Workflows: Training Data Generation: Created binary classification labels (Positive: Flood / Negative: Non-Flood) based on historical event data and rigorous ground-truth surveys using Geodetic GPS. Feature Engineering: Extracted, normalized, and annotated 19 spatial features (predictors) including DEM-derived indices, NDVI, and rainfall intensity for model input. Data Splitting & Validation: Structured datasets into Training (70%) and Testing (30%) subsets. Performed K-Fold Cross-Validation to ensure model robustness and prevent overfitting. Tools: Python (Scikit-learn, Pandas), ArcGIS Pro (for spatial labeling), and Google Colab.

Managed the end-to-end creation of a labeled geospatial dataset used to train and validate Supervised Machine Learning models (Random Forest, SVM, and XGBoost) for flood risk prediction. Key AI/ML Workflows: Training Data Generation: Created binary classification labels (Positive: Flood / Negative: Non-Flood) based on historical event data and rigorous ground-truth surveys using Geodetic GPS. Feature Engineering: Extracted, normalized, and annotated 19 spatial features (predictors) including DEM-derived indices, NDVI, and rainfall intensity for model input. Data Splitting & Validation: Structured datasets into Training (70%) and Testing (30%) subsets. Performed K-Fold Cross-Validation to ensure model robustness and prevent overfitting. Tools: Python (Scikit-learn, Pandas), ArcGIS Pro (for spatial labeling), and Google Colab.

2024 - 2025

Geospatial Object Detection for Urban Infrastructure & Risk Analysis

OtherGeospatial Tiled ImageryObject DetectionMapping
Developed a high-resolution Object Detection dataset used to assess urban vulnerability features (buildings, road networks, and drainage systems) for a flood risk model in the Tallo River Basin. Key AI/ML Workflows: Annotation (Bounding Box & Polygon): Annotated over 2,000+ urban infrastructure instances from high-resolution Drone/UAV imagery to train detection algorithms for urban density analysis. Data Augmentation: Managed dataset preprocessing including tiling, rotation, and color correction to improve model generalization under different lighting conditions. Quality Assurance (QA): Performed rigorous Human-in-the-Loop (HITL) validation to correct false positives (e.g., distinguishing between permanent buildings and temporary shelters). Outcome: The structured data was used to calculate "Exposure" indices for the Frequency Ratio statistical model published in Indonesian Tropical Geospatial journal.

Developed a high-resolution Object Detection dataset used to assess urban vulnerability features (buildings, road networks, and drainage systems) for a flood risk model in the Tallo River Basin. Key AI/ML Workflows: Annotation (Bounding Box & Polygon): Annotated over 2,000+ urban infrastructure instances from high-resolution Drone/UAV imagery to train detection algorithms for urban density analysis. Data Augmentation: Managed dataset preprocessing including tiling, rotation, and color correction to improve model generalization under different lighting conditions. Quality Assurance (QA): Performed rigorous Human-in-the-Loop (HITL) validation to correct false positives (e.g., distinguishing between permanent buildings and temporary shelters). Outcome: The structured data was used to calculate "Exposure" indices for the Frequency Ratio statistical model published in Indonesian Tropical Geospatial journal.

2024 - 2024

Education

H

Hasanuddin University

Master's Program, Regional Planning and Development

Master's Program
2024 - 2026
H

Hasanuddin University

Bachelor's Degree, Forestry

Bachelor's Degree
2017 - 2023

Work History

H

Hasanuddin University

Practitioner Lecturer

Makassar
2025 - Present
C

Center for Regional Research and Development

GIS Technical Expert Assistant

Makassar
2025 - Present