Executive Profile

Spatial Data Scientist and Geoinformatics specialist with demonstrated expertise building end-to-end machine learning systems integrating statistical modelling, geospatial analytics, and real-time inference. Specialised in spatial statistics, probabilistic modelling, causal inference, and production ML architecture across crime analytics, customer intelligence, and electoral geospatial systems. Proven ability to engineer modular, scalable data pipelines that translate complex multi-source spatial data into actionable intelligence. Equally fluent in the technical and communicative dimensions of data science — from model explainability and SHAP-driven insights to stakeholder-facing dashboards and automated reporting. Seeking a Master's programme in Data Science or Machine Learning to deepen expertise in scalable AI systems, generative modelling, and decision-making under uncertainty. Unpublished emerging poet; selected for the inaugural Utkarsh Poetry Fellowship, reflecting a commitment to precision in both technical and creative expression.

Skills
Programming Languages

Python, SQL, R, Java, JavaScript, HTML/​CSS, Bash/​Shell

Data Engineering & Databases

Pandas, NumPy, ETL Pipeline Design, Relational DB Schema Design, Data Cleaning & Deduplication, PostgreSQL/​SQL

Development Practices

Git, GitHub, Modular Architecture, Documentation, Unit Testing (basics)

Productivity and Reporting

Automated PDF/​CSV Reporting, Google Sheets Integration, Jupyter Notebooks

Machine Learning & Statistics

Scikit-learn, Statsmodels, Lifetimes (BG/​NBD), PyTorch, SHAP, FAISS, K-Means, Random Forest, GWR, Poisson/​NB GLMs, ETS, SVM, Logistic Regression

Web Frameworks & APIs

FastAPI, Streamlit, Dash, RESTful API Design

Cloud & DevOps

Docker (basics), AWS (fundamentals), Google Cloud Platform (fundamentals)

Geospatial & GIS

GeoPandas, Shapely, OSMnx, Folium, Pyproj, Fiona, QGIS, Rasterio, Spatial Autocorrelation (Moran's I, Getis-Ord Gi*), KDE

Visualisation

Plotly, Folium, Dash, Matplotlib, Seaborn

Operating Systems

Linux (Ubuntu), Windows

Education

Earth Sciences and Geoinformatics

Bangalore University
08/2023 – Present | Bangalore, India
  • •Double major in Geography and Geoinformatics, and minor in Finance
  • •GPA: 3.96/4.00 (weighted); 3.90/4.00 (unweighted) | Top 10% of cohort
  • •Specialization: Spatial Statistics, Geostatistics, Remote Sensing, Python/​R Programming for GIS
  • •Relevant Coursework: Spatial & Data Analysis in GIS, Python and R Programming for Spatial Statistics, Artificial Intelligence, Surveying & GNSS, Spatial Statistics & Geostatistics, Remote Sensing & Image Processing, Surveying Techniques
  • •Academic Distinction: Attendee, National GIS conferences (KSTA, International GIS Forum, GeoVision 2025)
  • •Served as Returning Class Representative, coordinating academic activities and events for a cohort of 30+ students and collaborating with faculty and external partners during symposiums and cultural festivals

    • IBM Data Science Professional Certificate (12-course specialization)

    Coursera⁠
    09/2025 – 12/2025 | Bangalore, India
  • •Average Score: 94% | GPA: 4.00/4.00
  • •Capstone Project: SpaceX Falcon 9 First-Stage Landing Prediction - Predictive modeling (SVM, Classification Trees, Logistic Regression); Interactive Dash dashboard; Geospatial analysis (Folium)
    • Experience

    NitiVerse Foundation

    GIS and Data Analyst Intern⁠
    02/2026 – Present | Bangalore, India
  • •Built ward-level electoral analysis pipeline, as measured by processing ~396 wards and thousands of booths, by building a QGIS spatial workflow integrating booth intersections, vote weighting, and automated aggregation.
  • •Delivered scalable geospatial datasets for political analytics, as measured by automated ward-wise vote share, margin, and derived metrics, by designing GIS models that compute weighted booth contributions and aggregate them to ward level.
  • •Developed data infrastructure for reporting and dashboards, as measured by export-ready ward datasets and automated report generation, by structuring spatial outputs for Google Sheets, PDF ward reports, and interactive Streamlit/Dash visualisation pipelines.
    • Technical Projects

    Spatial analysis of agroforestry expansion into wildlife corridors in Chikkamagaluru's coffee-forest mosaic and its relationship with human-elephant conflict, combining satellite classification, forest change detection, corridor modelling, and hotspot analysis.

  • •Land Cover Classification: Sentinel-2 multispectral imagery (10m) classified using Random Forest into forest, agroforestry, agriculture, and built-up classes; NDVI, NDWI, and EVI indices used as engineered features
    • Naval Kishore
    1 / 3
  • •Forest Change Detection: Hansen Global Forest Watch tree cover loss layers (2010–2023) processed to map temporal deforestation at forest-agroforestry boundaries within the Bhadra Wildlife Sanctuary buffer zone
  • •Corridor Resistance Modelling: Resistance surface constructed from land cover, slope (SRTM DEM), and proximity to roads and settlements; least-cost path analysis used to identify remaining functional elephant movement corridors
  • •Conflict Hotspot Detection: Kernel Density Estimation applied to georeferenced HEC incident records (Karnataka Forest Department) to generate smoothed conflict intensity surfaces
  • •Integrated Overlay Analysis: Corridor resistance and KDE conflict intensity overlaid to identify spatial clusters where encroachment pressure, corridor degradation, and conflict frequency coincide
  • •Tech Stack: Python, Google Earth Engine, GeoPandas, Rasterio, Scikit-learn, Shapely, QGIS, Matplotlib
  • •Context: Developed in the context of the International Conference on Human-Elephant Conflict Management (Karnataka Forest Department)

    • Land Use Change Detection Dashboard⁠

    Geospatial Change Analysis System
    03/2026 – Present

    Full-stack geospatial platform for detecting, quantifying, and visualising land cover change between two time periods.

  • •Real data pipeline: Ingested MODIS MCD12Q1 (2013) and Dynamic World v1 (2022) GeoTIFF exports at 30 m resolution (3,637 × 4,614 px); identified 446,934 ha of land cover change across a 7-class unified scheme with automated CRS alignment and NoData masking
  • •10 statistical analyses: Change matrix, Markov chain (steady-state + mixing time), Cohen's Kappa, Moran's I spatial autocorrelation (I = computed, queen contiguity), FRAGSTATS-style landscape metrics, Shannon entropy + KL divergence, Pontius net/​swap decomposition, chi-square independence test, FAO compound annual rate of change, and vulnerability index (Low → Critical)
  • •GEE integration: Auto-selects dataset by year (Dynamic World ≥2015, MODIS 2001–2014, ESA WorldCover 2021); downloads at 30 m via getDownloadURL; Nominatim geocoding for place-name area queries
  • •Export engine: Automated multi-page PDF report (ReportLab), 8-section CSV data bundle, and tagged GeoTIFF change map
  • •Tech Stack: Python, FastAPI, Rasterio, NumPy, SciPy, scikit-learn, Matplotlib, ReportLab, Google Earth Engine API, HTML/​CSS/​JS, Chart.js; deployed on Render (backend) + GitHub Pages (frontend)

    • Modular geospatial machine learning pipeline integrating spatial data engineering, statistical modeling, spatial statistics, time series forecasting, and interactive visualization for crime risk mapping and hotspot detection.

  • •Crime dataset: ~8 million incidents (2001–Dec 2025) aggregated to 500-meter hexagonal grid with multi-source feature engineering (streetlight locations, CTA bus stops, city boundaries)
  • •Spatial Autocorrelation: Moran's I = 0.5063 (p=0.0000), indicating moderate positive spatial autocorrelation; local clustering detected via Getis-Ord Gi* hotspot analysis
  • •KDE Intensity: Kernel density estimation on hex centroids for smoothed crime intensity surface generation
  • •Spatial Data Engineering: GeoPandas, Shapely; hexagonal grid generation, multi-source spatial joins, CRS management (WGS84→projected), temporal feature extraction
  • •Poisson & Negative Binomial GLM for baseline risk (with overdispersion diagnostics); Random Forest regression (200+ trees) for non-linear pattern capture and feature interactions; Geographically Weighted Regression (GWR) with automatic fallback to Local Linear modeling for grids >6K cells
  • •Time Series: Monthly crime forecasting via Exponential Smoothing (ETS) with graceful degradation for insufficient historical depth
  • •Dashboard: Interactive Streamlit visualization with choropleths (observed & predicted), temporal filtering (hour-of-day, day-of-week), animated crime maps, PDF/​CSV export

    • Corporate RFM Analytics Suite⁠

    Customer Intelligence Platform - Production-Ready ML System
    06/2025 – 08/2025

    End-to-end customer intelligence platform integrating advanced segmentation, probabilistic lifetime value modeling, churn prediction, causal uplift analysis, and real-time inference.

  • •Processed 397,884 transaction records → 4,338 unique customers generating £8.91M revenue
  • •4 customer segments identified via K-Means clustering: Premium Loyalists (705 customers, £5.75M revenue, 12.4% avg churn), At-Risk/​Churn Likely (1,617 customers, £0.57M revenue, 99.6% churn probability), Emerging Customers, Bargain-Driven Low Value
  • •Average CLTV: £1,704.95; Overall churn probability: 19.8%
  • •Tech Stack: Python, Streamlit, FastAPI, scikit-learn, lifetimes, SHAP, FAISS, PyTorch, Plotly
  • •Modularity: 7 specialized modules (RFM utils, CLTV models, uplift, embeddings, explainability, campaign planner, realtime serving) following production ML architecture patterns

    • Interactive Streamlit application for geospatial buffer zone analysis and parcel-road intersection detection.

  • •Dual input modes: File upload (shapefile ZIP, GeoJSON) or OpenStreetMap road network fetch via OSMnx
  • •Select and compare multiple buffer distances (10m, 50m, 100m, 200m, 250m) simultaneously
  • •Automatic CRS detection and projection; dynamic UTM zone determination for metric buffering; parcel-road intersection detection
  • •Folium maps with layer control; alternative tile layers (CartoDB, Terrain, Toner); real-time result display
  • •Export: Zipped shapefiles for downstream GIS processing
  • •Validation: 100% success rate on tested polygon/​road datasets; robust geometry checking and error handling
  • •Tech Stack: Streamlit, GeoPandas, Folium, OSMnx, Shapely, Fiona, Pyproj
    • Research Interest

    A Generative Microclimate-Aware Landscape Design Engine Using Surrogate Modelling and Multi-Objective AI Optimisation

  • •Objective: Build computational framework integrating microclimate modeling with generative landscape design via surrogate models and multi-objective optimization
    • Courses
    10/2025 – 02/2026
    Naval Kishore
    2 / 3
  • •Methodology: High-resolution geospatial/​remote-sensing data → fast, data-driven surrogate models of microclimate behavior → multi-objective AI optimization (thermal comfort, shading, vegetation, spatial constraints) → decision-support engine for climate-responsive landscape configurations
  • •Publication Target: Peer-reviewed journal + conference proceedings
    • Awards

    UTKARSH Poetry Fellowship⁠

    Bangalore Poetry Festival and Neena Verma Foundation
    2026

    Selected as 1 of 6 emerging, unpublished poets from 350+ applications for the inaugural year. Completed intensive multi-month fellowship spanning workshops, seminars, and creative modules. Authored 10 long-form poems, 20 haiku/senryu, 13 short-form poems across themes of nature, philosophy, history, emotions, and memory. Poetry curated for publication in 2026 Chapbook Anthology (Atta Galatta, independent Indian Language publisher) alongside cohort members' work.

    Languages
    English — Native/Bilingual
    French — Basic
    Spanish — Basic
    Hindi — Conversational
    Kannada — Fluent

    Spatial Analysis with Python - From Basics to Applied GIS

    The Nature's Eye
    09/2025 – 09/2025 | Bangalore, India
    04/2025 – 05/2025

    Fundamentals of Graphic Design⁠

    California Institute of the Arts
    01/2025 – 03/2025
    Interests
    Emerging Poet & Writer⁠
    • •Published works: 12+ original poems (nature-based, philosophical, historical, emotional themes) on personal poetry platform since early 2023
    • •Fellowship recognition: Utkarsh Poetry Fellowship cohort member (inaugural 2025 class); poems selected for 2026 Chapbook Anthology with Atta Galatta
    • •Intellectual diversity: Demonstrates creative thinking, narrative communication, and intellectual breadth complementing technical expertise
    Certificates
    National Conference on Climate Resilience and Sustainable Development — Karnataka Science and Technology Academy
    National Conference on Sustainable Groundwater Management for Water Security — Karnataka Science and Technology Academy
    International Conference on Human-Elephant Conflict Management — Karnataka Forest Department
    UGIT's 11th International Conference on Climate Change: Environmental and Soil Sustainability - Advanced Geospatial Solutions — School of Geography and Geoinformatics, BU
    UGIT's 12th International Conference on Climate Change and Sustainable Development of Natural Resources - Applications of Geospatial Technologies — School of Geography and Geoinformatics, BU
    UGIT's 14th International Conference - GeoVision 2025: Geospatial Innovation for Climate Action, Disaster Resilience, and Environmental Sustainability — School of Geography and Geoinformatics, BU
    Organisations

    Countryside Estate Plantation

    Farm Hand
    03/2018 – Present | Western Ghats, India

    Agricultural land in the Western Ghats coffee-forest mosaic — direct field context for agroforestry and forest-edge research in Chikkamagaluru District.

    WWF - India

    Community Volunteer
    01/2021 – Present

    Personal Initiative

    Environmental Education Volunteer
    05/2023 – 07/2023 | Karnataka, India

    Kaalinga Centre for Rainforest Ecology

    Student Volunteer
    04/2023 – 04/2023 | Agumbe, India
    Naval Kishore
    3 / 3