▚▚▚ ABOUT
Environmental data scientist working at the intersection of remote sensing, ecological modeling, and environmental data systems in Sri Lanka. Builds end-to-end ML pipelines — species range-shift and habitat modeling, land cover classification, embedding-based habitat detection, and geospatial foundation model fine-tuning — backed by production-grade data infrastructure and delivered through clean, minimal interfaces rather than dense legacy research tooling.
Trained in Statistics and Operations Research. Communicates methodology publicly through open, reproducible write-ups. Currently seeking research assistantship and postgraduate opportunities applying ML and information design to ecological and environmental data problems.
▚▚▚ STATISTICS
|
River basins covered by the CAMELS-Africa hydrological dataset. |
Historical groundwater records backfilled into a live PostGIS database. |
Land cover classification accuracy on Chundikulam Lagoon wetlands. |
Cross-validated habitat-signal detection accuracy, beating baseline by 7 pts. |
▚▚▚ CASE STUDIES
|
First large-sample hydrology dataset for Africa: 216 basin attributes across 44 years of daily data for 12,000+ river basins. Interactive geospatial dashboard (React 19, TypeScript, MapLibre GL) running an in-browser DuckDB-WASM query engine, served via AWS S3/CloudFront. » |
|
End-to-end land cover classification pipeline in Google Earth Engine using Google's AlphaEarth satellite embeddings (64-band, 10m) and a Random Forest classifier across 9 classes. Diagnosed a temporal label-misalignment bug that raised wetland-class recall from 18% to 88%. 87% overall accuracy, 83% recall on mangrove habitat. » |
|
Kernel-density-estimation range-shift analysis over 3,620 GBIF occurrence records spanning 34 years, with bandwidth selected via 5-fold cross-validation. Correctly attributed a 90x record increase to eBird adoption rather than population change. » |
|
Tested whether 128-dimensional satellite embeddings encode habitat quality across 2,921 UK butterfly-monitoring sites. 76.1% cross-validated accuracy vs. 69.0% baseline, validated with a permutation test (p = 0.0099) and a geographic control. » |
▚▚▚ CURRENT WORK
|
Consultant — Digital Innovations for Water Secure Africa (DIWASA), Phase 2 International Water Management Institute (IWMI) · Jul 2026 – Present GUI-based water accounting dashboards for the WA+ framework across Ethiopia, Ghana, Burkina Faso, and Zambia. ML/AI forecasting of water balance indicators at basin, country, and continental scale. |
|
Research Associate — Jaffna Peninsula Water Programme International Water Management Institute (IWMI) · Oct 2025 – Jun 2026 Centralized PostGIS-backed database consolidating groundwater and water-quality data, ingesting a live SCADA API across 39 monitoring stations and 2.6M historical records. |
▚▚▚ INDEPENDENT PROJECTS
- Wellspring — offline-capable field data capture PWA with OCR digitization and live salinity mapping
Next.js · Fastify · PostGIS · MapLibre GL - Trace — offline camera-trap species-ID desktop app using MegaDetector and SpeciesNet
Tauri - FieldNote / Warbler — bioacoustic bird-monitoring pipeline demoed on a Central Asian Flyway stopover site
BirdNET · SQLite · React - Ticket Issuer Workforce Scheduling — MILP shift-scheduling model cutting overtime from ~12 to 3 hrs/week
PuLP
▚▚▚ STACK
▚▚▚ PUBLICATIONS & OPEN-SOURCE RESEARCH SOFTWARE
- Vithushan, N., Niranjan, C. M., & Ranawaka, T. P. (2025). "Level of Interest and Engagement in Astrophysics Among Undergraduate Students in Sri Lanka." Postgraduate Institute of Science (PGIS), University of Peradeniya. Institutional Repository — nationwide survey (n=309, 28 universities) on astrophysics education access in Sri Lanka.
- shp2geojson — Rust CLI for converting ESRI Shapefiles to GeoJSON. Published on crates.io.
- pubsum — terminal-based open-access academic literature search and summarization CLI powered by the OpenAlex and CrossRef APIs.
▚▚▚ EDUCATION & CERTIFICATIONS
B.Sc. (Honours) in Statistics and Operations Research — University of Peradeniya, Sri Lanka 2020 – 2025
Coursework: Artificial Intelligence, Neural Networks, Statistics, Algorithms, Data Structures, Operations Research, Multivariate Methods, Basic Biology
Intro to Machine Learning — Kaggle (2025) · Version Control — Meta (2024) · Intro to iOS App Development — Meta (2023) · Intro to Front-End Development — Meta (2023)
Treasurer, INIC · Organising Committee Member, Hill Country Youth Peace Summit
decoding.py@gmail.com | linkedin.com/in/maria-niranjan | orcid.org/0009-0002-3166-8911