~ $ whoami
Machine Learning Engineer · Computer Vision · Machala, Ecuador
I build computer vision systems that leave the lab — segmentation models on edge devices for shrimp farming, satellite imagery for mangrove monitoring, LiDAR for forest inventories.
I am an Information Technology student at Universidad Técnica de Machala (2022–2027) and a machine learning engineer at LarvIA, where I design, train and deploy vision models on edge devices to optimize shrimp farming across Ecuador and Latin America.
On the research side, I was a Machine Learning & Data Science Research Intern at CNPEM — the Brazilian Nanotechnology National Laboratory — working on unsupervised learning for high-density electrochemical sensor data, and I maintain the reproducibility code for a manuscript on individual tree detection in airborne LiDAR submitted to Forests (MDPI). I presented at the LatinX in CV undergraduate consortium at CVPR 2024, where I received the best presentation award.
Machala, Ecuador
Campinas, Brazil
Machala, Ecuador
Machala, Ecuador
Machala, Ecuador
Deep-learning tree detection in airborne LiDAR
Inventory-anchored evaluation of deep-learning individual tree detection in airborne laser scanning (ALS), with label-efficient cross-domain transfer in alpine forests. Reproducibility code for a manuscript submitted to Forests (MDPI).
BLE access control with on-device face recognition
Intelligent Control Access and Security System for UTMACH classrooms. End to end: ESP32 (NimBLE) firmware, native Android & iOS apps with on-device face recognition (ML Kit + TFLite) and biometric auth, and a FastAPI REST backend.
Mangrove ecosystem mapping from satellite imagery
U²-Net model for mangrove segmentation using Landsat 7, 8 and Sentinel-2 imagery. CFMASK preprocessing to remove clouds and shadows; optimized for multispectral segmentation with RGB, NIR, SWIR-1 and SWIR-2 bands.
Small-object detection in urban aerial imagery
Improving small-object detection with YOLO and Slicing Aided Hyper Inference, evaluated on VisDrone with real mAP metrics. Talk & demo for FLISOL 2026.
Forests (MDPI) · manuscript submitted
Inventory-anchored evaluation of deep-learning tree detection in ALS point clouds, with label-efficient cross-domain transfer in alpine forests. Public reproducibility code (PyTorch, PointNet++).
CNPEM · Campinas, Brazil · 2026
Unsupervised clustering of electrochemical reaction signals and multi-label classification of sensor-related patterns, during a research internship at the Brazilian Nanotechnology National Laboratory.
Seattle, USA · best presentation award
Selected extended abstract, presented at the undergraduate consortium with a travel scholarship from the Computer Vision Foundation.
Universidad Técnica de Machala · Machala, Ecuador
ESPOL · Guayaquil, Ecuador
FGV EMAp · Rio de Janeiro, Brazil
Participation scholarship · Santiago, Chile
$ ./get_in_touch.sh
Open to research collaborations, ML engineering work, and speaking at community events — especially anything involving computer vision in the real world.