Agentic AI
Developer
Building multi-agent systems, RAG pipelines, and geospatial automation workflows with Python. Based in India — available remotely.
Education
Expertise
AI & Data
Geospatial
Frameworks
Tooling
Career
- Georeferencing and boundary creation for multi-country data frames.
- Automated GIS tasks with Python to process and analyse vector data at scale.
- Built PyQGIS & GeoPandas scripts to clean, validate, and modify shapefiles.
- Data validation and quality checks to maintain error-free boundary maps.
- Automated merging, attribute updates, and geometry corrections from multiple sources.
- Python training for data & GIS professionals (Pandas, Matplotlib, Scikit-learn).
- Mentored 4+ batches (3–4 months each) on automation and data processing.
- Designed hands-on exercises and workflow optimisation solutions.
Projects
Multi-agent QA workflow using LangGraph that transforms raw user stories into prioritized Gherkin test suites. Three specialized agents (Requirement Analyzer, Test Generator, Test Reviewer) with autonomous Tavily tool-calling for grounded citations, streaming UX, and LangSmith observability.
AI-powered system to query any geospatial dataset (GeoJSON, SHP, KML) using natural language. Uses LLM tool-calling instead of plain RAG — no hardcoded schema — with multi-step execution, interactive PyDeck maps, and a built-in EDA dashboard.
Multi-agent system for personalised fitness and diet planning — user profiling, RAG, expert agents, and automated PDF report generation via ReportLab.
High-performance pipeline converting large shapefiles into structured KML outputs using GeoPandas + Shapely, with multiprocessing and custom XML formatting.
Land cover classification with SVM, Random Forest, Decision Tree, and XGBoost. Full preprocessing, feature engineering, accuracy metrics, and classification maps.
Unsupervised pipeline for Sentinel-2B imagery using K-Means + PCA. Clustering evaluation and visual insights for coastal land analysis.
Logistic regression model for flood risk — preprocessing, training, evaluation, and visualisations to identify high-risk zones.
Automation pipeline to compute NDVI from Sentinel imagery (.img, .tiff) with Rasterio and NumPy. Outputs GeoTIFFs for scalable vegetation health monitoring.