Atmospheric & Climate Data Specialist at Wind Pioneers — I build the data infrastructure that wind energy decisions rest on.
Working in atmospheric and climate data since 2021. Most of my work sits between the science and the software: reanalysis pipelines that have to be right, hazard models that have to be defensible, and tooling that turns a half-day GIS chore into a one-minute command. I care about systems that refuse to ship quietly broken data.
Reanalysis at scale. ERA5 and MERRA-2 wind timeseries served from a BigQuery-backed warehouse and cloud-native zarr stores, with acquisition pipelines running from CDS and NASA GES DISC through to Google Cloud Storage.
Correctness as a feature. Completeness checks that raise rather than return gaps, deterministic nearest-node lookup that survives the antimeridian and the poles, dependency constraints pinned against real read failures rather than guesses.
Hazard modelling. Typhoon wind fields from IBTrACS best-track data, extreme wind return periods for offshore sites, and the validation work that decides whether a number is usable.
🌬️ WindSync / Reanalysis SDK Python library for ERA5 and MERRA-2 wind speed and direction timeseries. Three interchangeable providers — BigQuery, GCS zarr, local netCDF — behind one API, plus a CLI and a packaged desktop app. Refuses incomplete series by default instead of silently shipping gaps. Cut reanalysis processing time by 75% and removed 95% of the manual handling.
🗺️ NEWA Map Extraction Pulls 50 m New European Wind Atlas maps for a project area straight from a site KMZ or bounding box and writes them in the form the GIS blend expects. Minutes instead of half a day — and it checks the output rather than trusting it. The high-resolution map matters: over one Bulgarian project area the 3 km map flattened the ridges the sites actually sit on, putting 24 of 42 sites above 6.0 m/s instead of 10.
🌪️ Extremes Toolkit Extreme wind statistics for offshore sites. Holland vortex typhoon wind fields driven by IBTrACS tracks, return-period estimation via EWTS, MIS and block maxima, and a desktop runner so the analysts using it don't need a terminal.
🛰️ sitestack Stacks Earth-observation land surface temperature and precipitation timeseries for a point — twelve products across nine sources and five access mechanisms, normalised into one long-format schema with per-product zarr cubes on their native grids. Adding a new site is one YAML file.
🌀 Holland Wind Field Model — numerical cyclone/typhoon wind profiles for wind farm risk assessment. 📦 NetCDF Maker — R Shiny app converting raw CSV sensor data into standardised NetCDF. 🌦️ High-Resolution Atmospheric Dataset — 18% forecast precision gain via deep-learning bias correction and spatial downscaling.
Languages · Python, R, SQL, MATLAB, Swift
Climate & geospatial data · NetCDF, GRIB, Zarr, HDF5 · ERA5, MERRA-2, ECMWF, GFS, NEWA, IBTrACS · xarray, CDO, NCO, GDAL, rasterio, QGIS, ArcGIS
Modelling · WRF, HYSPLIT, Holland vortex · extreme value statistics (EWTS, MIS, block maxima) · statistical downscaling · ML bias correction (LSTM, CNN) · Windographer
Platform & engineering · Google Cloud (BigQuery, GCS, CI/CD) · Docker, Poetry, pytest, ruff, pre-commit · REST APIs, Plumber, Swagger · Git
Visualization · matplotlib, Plotly, Leaflet, ggplot2
M.Sc. Atmosphere & Ocean Sciences — Indian Institute of Technology Bhubaneswar Dynamics, numerical modelling and ocean–atmosphere coupling; the grounding behind the hazard and reanalysis work above.
B.Sc. Geology — St. Xavier's College, Ranchi Earth systems, field mapping and remote sensing — where the geospatial side started.
Extreme rainfall variability & intraseasonal dynamics Characterising how monsoon rainfall extremes vary on intraseasonal timescales, and what drives the swings.
Kalman filtering on HPC Sequential state estimation applied to atmospheric data, run on high-performance computing infrastructure.
India Meteorological Department (IMD) — field training at India's national weather service National Atmospheric Research Laboratory (NARL), Gadanki — India's atmospheric radar and lidar facility Indian National Centre for Ocean Information Services (INCOIS), Hyderabad — operational ocean forecasting and marine advisories
Scientifically sound, production-ready climate intelligence for real-world decisions.



