Abir Raihan

July 2024 — Present

Developing production geospatial and agricultural data systems that connect cloud processing, spatial analytics and operational decision support.

• Design and maintain automated ingestion and transformation workflows for large agricultural datasets (field boundaries, planting, harvest and yield, fertilizer application, soil and weather), turning provider-specific data into standardized, analysis-ready products and mapping crop and variety names to controlled definitions while preserving the raw values.
• Own the John Deere integration end to end: technical discovery and due diligence, customer authorization, webhook-driven processing, validation, standardization, failure reporting and delivery to BigQuery.
• Build event-driven processing on Cloud Run, Cloud Storage, Pub/Sub and Cloud Tasks, with scheduled recovery, replacing manual, SFTP and desktop-era workflows with observable, repeatable cloud processes.
• Engineer reliable synchronization from operational PostgreSQL/PostGIS into BigQuery using a transactional-outbox pattern, dbt staging and marts, batching, monitoring and automated recovery.
• Designed workload-based routing so standard datasets stay on a simple Python path while very large ones can move to Spark, Apache Sedona and Dataproc Serverless, balancing performance against cloud cost.
• Built the Model Context Protocol (MCP) service, in FastAPI on Cloud Run, that gives AI applications controlled access to the analytical platform, keeping trusted analytics and predictive models separate from LLM interpretation and explanation. The wider LLM-agent work is a team effort.
• Own production monitoring and troubleshooting, and ship through Terraform-managed infrastructure and review-gated CI/CD.
• Partner with agronomy, analytics, account-management and engineering teams to turn operational problems into scalable data products.

April 2020 — June 2024

Built and scaled automated data, remote-sensing and analytical workflows for precision-agriculture operations.

• Led advanced geospatial and agronomic analysis that turned soil, yield, planting, application and management-zone history into field-level recommendations.
• Replaced repetitive desktop-GIS and agronomic-software workflows with Python automation, taking fertilizer and prescription preparation from half a day or more to roughly 15–30 minutes and making it repeatable across analysts.
• Developed variable-rate fertilizer recommendation logic that compares current soil-test values with the nutrient requirements for a target yield, using regression against historical yield, and validated it on trial plots and prior-year results.
• Interpolated soil properties into continuous surfaces with IDW and ordinary kriging (Gaussian variogram), and evaluated regression kriging and satellite-imagery relationships for nutrient and productivity analysis.
• Built variable-rate planting prescriptions that estimate zone-level yield potential from soil samples and multi-year yield history and assign differentiated seeding rates, with an analyst review step before release.
• Automated geometry preparation for legacy John Deere monitors by partitioning field boundaries to their six-mile constraint, cutting a one-to-two-day GIS task to roughly 7–8 minutes and letting growers keep working equipment in service.
• Mentored teammates on geospatial analysis, data quality and automation, and helped move the team from manual GIS processing toward reusable, programmable workflows.

January 2019 — March 2020

Delivered spatial processing, field-level analysis and GIS workflows supporting agricultural services and planning.

• Managed, validated and maintained field boundaries, soil-sample datasets and related spatial and tabular data from multiple platforms and providers.
• Produced maps, spatial reports and field-level outputs for agronomic planning and customer communication, using overlays, spatial joins, buffering and zone processing.
• Linked laboratory soil results to sample locations and prepared boundaries, samples and management zones for variable-rate projects.
• Wrote early Python scripts and internal tools to automate repetitive GIS and desktop-software tasks, the start of an automation practice that later grew into cloud engineering.

January 2017 — December 2018

Applied GIS, remote sensing and spatial analysis to plant, soil and agricultural research.

• Coordinated same-day collection of in-situ soil samples, UAV multispectral and thermal imagery and satellite overpass data so the datasets were temporally aligned.
• Processed multi-source remote-sensing and field data in Python and R, deriving image, vegetation and thermal indices and evaluating their relationship with measured soil moisture.
• Validated remote-sensing estimates against field soil measurements and reported the results in a thesis and a conference presentation, and co-authored a published paper on field-scale soil variability.

GIS, Remote Sensing, Precision Agriculture, Satellite and UAV Imagery, Soil Science and Spatial Analysis.

Coursework Geospatial Analysis and Interpretation; Precision Agriculture

Thesis Surface Soil Moisture Estimation Using Unmanned Aerial System and Satellite Images. link ↗

Land-Use Analysis, GIS, Spatial Planning and Urban-Growth Modeling.

Coursework Geographic Information Systems I and II; Remote Sensing I and II

Thesis Measuring Urban Sprawl and Its Effect on Public Service Facility Around Khulna City: A Geographic Information System and Satellite Remote Sensing Approach.