Developing open-source super-resolution for satellite imagery. Making high-res Earth observation accessible through AI-enhanced Sentinel-2/Landsat data.

About Me

I'm Bilel Khlaifia,a Geospatial AI Engineer developer specializing in satellite image enhancement through deep learning. My mission is to democratize access to high-resolution Earth observation data by transforming freely available satellite imagery into enhanced quality that approaches commercial alternatives.

What I Do

I develop super-resolution algorithms that enhance Sentinel-2 (10m) and Landsat (30m) imagery to achieve 2-10x resolution improvements, making detailed Earth observation accessible to researchers, NGOs, and organizations that cannot afford expensive commercial satellite data.

Current Projects

Satellite Super-Resolution Pipeline
Open-source toolkit for enhancing multispectral satellite imagery using state-of-the-art deep learning models (ESRGAN, SRGAN, Vision Transformers).

Multi-Temporal Fusion Framework
Algorithms that intelligently combine multiple satellite passes to improve spatial resolution while preserving spectral and temporal accuracy.

Educational Aerospace Platforms
Developing affordable UAV and CanSat designs for remote sensing education and research, with complete documentation for universities and makers.

Why This Matters

Commercial satellite imagery can cost thousands of dollars per scene. My tools enable:

• Researchers in developing countries to access detailed imagery for environmental monitoring
• Small farmers to monitor crop health without expensive subscriptions
• Disaster response teams to quickly assess damage using free data
• Students to learn remote sensing with professional-grade tools

Technical Approach

• Training specialized neural networks on paired low/high resolution satellite data
• Optimizing for multispectral imagery preservation
• Ensuring geographic and radiometric accuracy
• Building scalable processing pipelines using GDAL, Rasterio, and PyTorch
• Validating against commercial imagery (SSIM: 0.85+, PSNR: 30+ dB)

What Sponsorship Enables

Your support directly funds:

• GPU compute time for training larger, more accurate models
• Validation datasets from commercial providers
• Open-source development and maintenance
• Documentation and tutorial creation
• Community support and education
• Hardware components for aerospace prototypes

2026 Goals

• Release production-ready super-resolution models for Sentinel-2
• Achieve consistent 3-5m resolution from 10m source imagery
• Process 100,000+ sq km of imagery for the community
• Launch educational CanSat program with 5 universities
• Create comprehensive remote sensing course materials

Community First

All core tools remain open-source and free. Sponsorship enables sustainable development and ensures these resources remain available to everyone, especially those who need them most.

Featured Work

• 🛰️ Sentinel-2 Enhancement Pipeline (2-3x resolution improvement)
• 🤖 Multi-temporal Fusion Algorithm (combining 10+ scenes)
• 🚁 Open UAV Design for Multispectral Imaging
• 📡 Educational CanSat Platform with Full Telemetry

Contact & Links

• Email: khlaifiabilel@icloud.com
• LinkedIn: linkedin.com/in/khlaifiabilel