A comprehensive CLI tool that extracts text from images and documents using multiple cutting-edge AI providers. Choose from privacy-focused local models like SmolVLM and LLaVA, or leverage high-accuracy cloud-based solutions like OpenAI GPT-4o. Features one-command setup with Pixi, custom extraction prompts, batch processing capabilities, and support for both local files and web URLs. Perfect for document digitization, receipt processing, research paper analysis, and enterprise document workflows with maximum flexibility and data privacy control.
I design and maintain robust data delivery environments that convert raw geospatial and remote sensing inputs into operational intelligence. From scalable SAR raster pipelines to collaborative dashboards and feature‑rich fusion outputs.
A showcase of hackathons that amplified my expertise in Artificial Intelligence, machine learning, geospatial intelligence, LiDAR, and entrepreneurship—focusing on real-world data challenges and interdisciplinary teamwork in forestry and environmental domains.
The transferable capabilities bridging Earth science, data science, and strategic decision-making.
My work in SAR focuses on advancing techniques like polarimetric decomposition, calibration, and 3D tomographic imaging to better characterize land and forest structure. Leveraging datasets from platforms such as UAVSAR and TerraSAR-X, I’ve investigated frequency-dependent scattering, refined PolSAR calibration workflows, and developed vertical profiling methods using PolSAR tomography. These efforts support applications in ecosystem monitoring, geophysical modeling, and the interpretation of complex surface features.
My work with InSAR focuses on detecting and analyzing ground displacement using radar remote sensing techniques such as DInSAR and PSInSAR. I’ve developed and implemented workflows for processing satellite SAR data, enabling accurate monitoring of subsidence, tectonic motion, and anthropogenic changes. These efforts have supported environmental and geophysical studies across varied terrains using Sentinel-1 and TerraSAR-X datasets.