Skip to content
View Ranjit-Saha's full-sized avatar

Block or report Ranjit-Saha

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
Ranjit-Saha/README.md

Ranjit Saha

Geo-Computational Product Engineer | Spatial Data & ML Pipeline Architect

Architecting full-stack spatial intelligence pipelines and deployment-ready machine learning systems to bridge the gap between physical-world anomalies and scalable software.

LinkedIn Email

🌍 Engineering Focus

I specialize in the backend orchestration of Earth Observation data and the deployment of inference-only machine learning architectures. My work focuses on transitioning theoretical models into production-grade, failure-aware pipelines capable of handling massive spatial telemetry and enforcing strict data integrity.

📁 Flagship Projects

🛰️ Agri-Sentry-core

  • Architecture: A climate-resilient spatial backend designed to eliminate "Basis Risk" in parametric insurance by detecting flood anomalies (-18.0 dB threshold) through heavy monsoon cloud cover.
  • Infrastructure: PostGIS (RLS & GiST) · Google Earth Engine API · Sentinel-1 SAR · GeoPandas · Vectorized Processing (NumPy)

🏥 Health Insurance Cost Predictor

  • Architecture: A deployment-ready regression inference pipeline featuring segmented modeling for different demographics, ensuring strict separation of training and inference workflows to prevent runtime data leakage.
  • Infrastructure: Scikit-Learn · Joblib (Serialized Assets) · Streamlit · Python OOP

🏔️ Terra-Valuation Engine (Aravalli Audit)

  • Architecture: Geospatial forensic system built to quantify environmental compliance and ecological liability over a 34,000+ hectare mining region.
  • Infrastructure: GEE API · JAXA ALOS Radar · Sentinel-2 Multispectral · Streamlit UI

🚗 Decoupled CV Microservice

  • Architecture: A distributed Computer Vision microservice for automated vehicle damage assessment, operating with zero-disk I/O to prevent server bottlenecks under concurrent load.
  • Infrastructure: PyTorch (ResNet50) · FastAPI Asynchronous Engine · Docker Containerization

⚡ Infrastructure & Pipeline Capabilities

  • Cloud-Native Ingestion: Designed API-driven handshakes with Google Earth Engine to extract, filter, and process multi-year satellite backscatter histories.
  • Secure Spatial Vaults: Architected idempotent PostGIS databases with Row-Level Security (RLS) for multi-tenant data isolation and precise 3m inner-core boundary buffering.
  • Deployment-Oriented ML: Built segmented regression architectures utilizing serialized machine learning assets (joblib) and pre-fitted standard scalers to guarantee inference environment stability.
  • Compute Efficiency: Engineered NumPy-vectorized raster processing pipelines to achieve massive execution speedups over standard iterators.

🔧 Core Technical Stack

Python PostGIS PyTorch FastAPI Docker Streamlit

  • Earth Observation: Sentinel-1 (SAR) · Sentinel-2 · Google Earth Engine API
  • Spatial Engineering: PostGIS · PostgreSQL · GeoPandas · Rasterio · GDAL · Shapely
  • Machine Learning & AI: PyTorch · Scikit-Learn · Segmented ML Modeling · NumPy (Vectorization)
  • Backend Infrastructure: FastAPI · Docker · Microservices · Joblib Serialization

📊 Engineering Activity

Stats Streak

Pinned Loading

  1. vehicle-damage-detection-microservice vehicle-damage-detection-microservice Public

    A decoupled Computer Vision microservice featuring an asynchronous FastAPI inference engine (ResNet50) and a lightweight Streamlit client for vehicle damage assessment.

    Python

  2. agri-sentry-core agri-sentry-core Public

    Production-grade spatial backend integrating Sentinel-1 SAR, PostGIS, and Google Earth Engine for automated flood detection and parametric insurance.

    Jupyter Notebook

  3. aravalli-forensic-audit aravalli-forensic-audit Public

    Geospatial forensic audit engine quantifying hidden ecological liability in mining belts using Google Earth Engine, JAXA ALOS (30m), and multi-temporal Sentinel-2 imagery.

    JavaScript

  4. insurance-premium-predictor-ml insurance-premium-predictor-ml Public

    A deployment-ready regression pipeline predicting health insurance costs, featuring segmented modeling and an inference-only Streamlit architecture to prevent runtime data leakage.

    Python