PB Puspendu Biswas Paul Geospatial AI · SAR · Risk Mapping

Remote Sensing & Geospatial AI Researcher · Hangzhou, China

Turning Earth observation data into environmental insight.

I combine remote sensing, GIS, Python automation, and AI/ML workflows to transform satellite, terrain, and climate data into practical outputs for research, planning, risk analysis, and decision support.

Contact me GitHub
CSC Scholarship IELTS 7.5 Tools Python · GEE · ArcGIS · PyTorch
SAR flood mappingAttention U-NetDeepLabV3+DEM validationFwDET depth mappingPython automationArcGIS ProGoogle Earth EngineSNAPRisk communication

Profile

Remote sensing, AI, and GIS as one connected workflow.

My work sits between remote sensing research and practical decision support. I integrate Sentinel-1 SAR, terrain, climate, and GIS datasets using ArcGIS Pro, Google Earth Engine, SNAP, Python, and PyTorch to produce flood-risk, environmental-risk, and local planning outputs.

My work follows the full analytical chain: data cleaning → feature-stack preparation → model comparison → environmental-risk products → local-scale interpretation → clear maps and reports for research, planning, and decision support.

Experience

Research and professional path

Built around environmental risk, flood mapping, and decision-support geospatial analysis.

May 2025 — Present

Scientific Officer Part-time

International Innovation Institute, Beihang University · Hangzhou, China

Developed local-scale flood-information workflows using Sentinel-1 SAR, DEMs, Python, ArcGIS Pro, Google Earth Engine, and PyTorch. Compared Attention U-Net and DeepLabV3+ flood-segmentation workflows and translated outputs into parcel/building/owner-level impact layers.

Apr 2025 — Present

Research Member Voluntary

Hydro-Climate & Ocean Research Centre · Dhaka, Bangladesh

Support GIS mapping, spatial analysis, coastal-hazard thinking, GNSS-R applications, climate-induced risk analysis, and early-warning research planning.

Jan 2024 — Aug 2024

Geospatial Officer

Science Connect Ltd. · Dhaka, Bangladesh

Worked on land-elevation assessment and low-lying flood-susceptibility projects. Prepared geospatial datasets, spatial layers, terrain analysis, maps, and analytical outputs for environmental-risk reporting.

Aug 2023 — Dec 2023

Disaster Mitigation & Spatial Data Intern

RIMES Bangladesh Satellite Office · Dhaka, Bangladesh

Supported disaster-preparedness and multi-hazard early-warning workflows through GIS data preparation, risk-analysis assistance, and stakeholder-oriented mapping outputs.

Mar 2021 — Jan 2023

Research Assistant

Sylhet Agricultural University · Sylhet, Bangladesh

Managed drought-index datasets, statistical analysis, research documentation, laboratory support, and early environmental modelling work.

Expected June 2026

M.S. Space Technology Applications

Beihang University, China

Research focus: AI-enabled flood information, Sentinel-1 SAR flood mapping, DEM validation, multi-sensor hydrological analysis, and parcel-level flood-risk interpretation.
2018 — 2023

B.Sc. Agricultural Engineering

Sylhet Agricultural University, Bangladesh

Final thesis: Prediction of Drought Index Using Advanced Fuzzy-Logic Model.
Award

Chinese Government Scholarship

CSC Scholarship, 2024

Graduate study support for research in space technology applications and remote sensing.

Skill stack

Technical capabilities for geospatial research, automation, and decision support

Geospatial & Remote Sensing

ArcGIS Pro, QGIS, Google Earth Engine, SNAP/Sentinel-1 Toolbox, Sentinel-1 SAR, optical imagery, raster/vector analysis, DEM validation, exposure mapping, geodatabase design.

Python Geospatial Automation

Raster/vector batch processing, SAR feature-stack preparation, geospatial data cleaning, spatial overlay automation, reproducible notebook workflows, pandas, GeoPandas, Rasterio, Xarray, NumPy.

Machine Learning & AI

PyTorch, CNN concepts, Attention U-Net, DeepLabV3+, semantic segmentation, image classification, fuzzy-logic modelling, feature engineering, model comparison.

Climate & Environmental Risk

Flood/drought indicators, rainfall-event interpretation, terrain/elevation analysis, environmental variable integration, early-warning and risk-map production.

Research & Reporting

Data-quality checks, sampling design support, accuracy assessment, statistical tables, publication figures, technical reports, manuscripts, conference materials.

AI-Assisted Work Efficiency

AI-assisted coding, debugging, literature screening, technical writing, data-quality checking, workflow planning, and research-output polishing for remote sensing/GIS analysis.

Selected projects

Selected work across Earth observation, environmental risk, and applied AI

Projects are organized by analytical problem, method, dataset, and practical output.

Flood Intelligence

Flood extent and depth mapping in Bangladesh

Multi-stage workflow combining Sentinel-1 SAR, Attention U-Net, DEM assessment, FwDET depth estimation, and parcel-level GIS impact interpretation.

  • SAR before/after feature stacks
  • Deep learning flood segmentation
  • Depth and exposure translation
Terrain Reliability

DEM validation for Jamuna floodplain

Assessed open-access DEMs using water- and land-domain reference data, bootstrapped uncertainty, and decision-oriented ranking for hydrological suitability.

  • DEM error metrics
  • Water/land validation
  • Flood-depth suitability support
Hydroclimate

GRACE-based drought-storage dynamics

Processed GRACE/GRACE-FO and climate datasets to analyze terrestrial-water storage behavior using machine-learning classification logic.

  • Storage-drought response
  • Climate dataset processing
  • Hydrological pattern analysis
AI Dataset

BD freshwater fish image dataset

Contributed to AI-powered automatic fish species classification/detection work supporting smart aquaculture and environmental data applications.

  • Image dataset contribution
  • Classification/detection support
  • Smart aquaculture context

Publications

Research outputs and manuscripts

Proceeding · IAASPACE 2025

Optimizing Earth-Moon Transfer and Cislunar Navigation: Integrating Low-Energy Trajectories, AI Techniques and GNSS-R Technologies

DOI: 10.48550/arXiv.2511.03173
Published

Impact of water pollution on physio-chemical properties for fish habitat in the Surma River, Sylhet city, Bangladesh

DOI: 10.52293/WES.2025.1.1042
Published · Data in Brief

BD-Freshwater-Fish: An Image Dataset from Bangladesh for AI-Powered Automatic Fish Species Classification and Detection toward Smart Aquaculture

DOI: 10.1016/j.dib.2024.111132
Ongoing manuscript

Hydrologic suitability assessment of global DEMs in the Jamuna floodplain using remote-sensing-driven water and river metrics

Manuscript in progress.

Professional focus

Connecting satellite data, modelling, and decision support.

My work combines remote sensing, geospatial automation, AI/ML, and environmental analysis to create reproducible workflows and clear outputs for research, planning, early warning, and operational decision-making.

Contact

Open to geospatial AI, remote sensing R&D, environmental analytics, and decision-support roles.

I am interested in roles that connect satellite data, AI/ML, Python automation, and applied environmental or risk analysis.

References

Prof. Sheikh Tawhidul Islam — International Innovation Institute, Beihang University.

Dr. Mohan Kumar Das — Executive Director, Hydro-Climate and Ocean Centre, Dhaka, Bangladesh.