Computer Vision Specialist

EchoGlobal is developing an agri-fintech platform that leverages advanced crop monitoring and data analytics to support lending and farm management decisions. The company seeks a Computer Vision Specialist to design and deploy semantic segmentation models for orchard structure detection from high-resolution satellite imagery, powering a global agrifintech MVP.

• Design and implement semantic segmentation models (DeepLabv3+, U-Net, or similar) trained on 15 cm RGB satellite imagery tiles
• Build preprocessing pipelines for raster tiling, dataset preparation, and georeferenced data processing using GDAL and Rasterio
• Implement augmentation strategies, loss functions, and hyperparameter tuning for structured agricultural scenes
• Perform sliding-window inference optimization for large-area batch processing
• Support mask post-processing including morphology, skeletonization, and polygon extraction for GIS compatibility
• Define and track segmentation metrics (IoU, precision/recall) and perform field-level validation
• Collaborate with Technical Lead (GIS & Remote Sensing), Labeling Specialist, and Backend/DevOps teams
• Improve model performance using pseudo-labeling and semi-supervised techniques

• 3+ years of professional experience in Computer Vision
• Strong proficiency in PyTorch (preferred) or TensorFlow
• Hands-on experience with semantic segmentation architectures
• Experience working with high-resolution imagery (satellite, drone, or aerial)
• Demonstrated experience with semi-supervised learning techniques
• Experience with geospatial data and coordinate systems/projections
• Preferred: Agricultural or remote sensing project background
• Preferred: Exposure to transformer-based segmentation models (e.g., SegFormer)
• Preferred: Experience integrating ML pipelines into backend systems
• Preferred: Knowledge of PostGIS or spatial databases

Schedule: Monday-Friday 9 AM to 5 PM (Uzbekistan time)

Work Environment: Cloud GPU infrastructure (RTX 3090 / T4 / A100 class), satellite imagery processing, technical stack includes PyTorch, GDAL/Rasterio, PostGIS, and LabelMe annotation tools

Note: Client requires work-time overlap within their time zone