Skip to content

Bidhyananda Yadav

Geocomputational Scientist  ·  Spatial Data Engineer  ·  ML/AI Researcher

Seven years designing TB-scale Earth observation pipelines, ML/AI workflows, and HPC infrastructure for NASA and NSF programmes at The Ohio State University.

Engineering Notes Workflows Blog

What I Build

  • Geospatial & Remote Sensing

    Processing optical, SAR, LiDAR, and hyperspectral satellite data at TB scale. ETL pipelines for continental coverage built on Xarray, Rioxarray, GDAL, and GeoPandas.

  • Scientific ML & AI Systems

    Doctoral research in applied ML: ensemble methods (Random Forest, Gradient Boosting, Extra Trees), regularized regression, and geospatial feature engineering across 1,200+ USGS gages at continental scale. Extended since into deep learning integration, remote-sensing data pipelines, and AI application support for research teams.

  • HPC & Cloud Infrastructure

    Distributed SLURM cluster workflows and AWS-native pipelines (Lambda, S3, containers). Designed for reproducibility and seamless HPC-to-cloud migration at scale.

  • NASA SWOT Satellite Mission

    Co-led remote sensing algorithms for global river discharge and lake dynamics. Led the initial prototype of worldwide SWOT analytics dashboard (swotviz.cuahsi.io) serving international hydrology teams.

    NASA Grant · $925,353 · 2022–2025

  • Continental Snow Water Equivalent

    Architected a 20-year daily SWE data product for North America at 0.01° resolution (~30 TB). Fused MODIS satellite observations with LISF land surface model outputs. Implemented on NASA Discover HPC clusters using Julia and JuMP. Open-source pipeline.

    NASA Grant · 2022–2025

  • ICESat-2 & ArcticDEM Elevation Toolkit

    Python toolkit to download and process ICESat-2 laser altimetry data across Antarctica and the Arctic, correcting systematic elevation biases in high-resolution DEMs (Data products: ArcticDEM, REMA, WorldDEM). A multi-agency collaborative work with Universities and Federal Agencies.

  • Greenland Glacier Velocity

    Satellite pipeline to map ice velocity across Greenland outlet glaciers using Sentinel-2 (A/B/C) and Landsat imagery. Open-source pipeline with data delivered to NSIDC DAAC. Published in The Cryosphere.

  • Fluid Earth Viewer

    An interactive web application visualizing global atmospheric and ocean conditions for public science communication. Implemented climate metric aggregations (monthly, seasonal, annual) and derived data products. Later adopted by the Permafrost Discovery Gateway. fever.byrd.osu.edu

    NSF Grant

  • Deep Learning for ArcticDEM Hydrology

    Built satellite-data pipelines, preprocessing workflows, and downstream integration layers for PyTorch-based water classification research using Sentinel-2 and commercial high-resolution imagery. Published in Journal of Hydrology and Remote Sensing of Environment.

Explore the Knowledge Base

  • Engineering Notes

    Evergreen references across three domains: AI & Machine Learning (transformers, LLM inference, fine-tuning), Geospatial & Remote Sensing (the Python stack, coordinate systems, sensor physics), and HPC & Infrastructure (distributed execution, on-premise model serving). Written as public reference material to help domain scientists build practical intuition for AI/ML workflows.

    Browse Engineering Notes

  • Workflows

    End-to-end Jupyter notebooks: the NASA PACE hyperspectral series, a satellite change detection pipeline, and applied NLP workflows.

    Browse Workflows

  • Blog

    Dated field notes on fast-moving tools and fields: running AI coding agents across a laptop and a shared HPC cluster, a 2026 map of computer-vision models for remote sensing, and other things that will look different in a year.

    Read the Blog