Weather Data Scientist
Job description
About the role
You will own the design and execution of numerical weather prediction experiments that underpin our core forecasting products for India. You will build and maintain ML-ready datasets that bridge observational gaps and support model training across data-sparse regions. You will benchmark multiscale forecasting systems against both reanalysis and real-world observations, with a focus on nowcasting and extreme weather events. You will run end-to-end cycling data assimilation and forecast loops at convection-permitting resolution for key Indian sub-regions. You will tailor models to renewable energy applications, specifically for solar and wind generation forecasting. You will work at the interface of physics-based models and machine learning, including hybrid physics-ML systems and emulator development. You will verify forecast performance using deterministic and probabilistic metrics to guide model improvements. You will communicate technical findings clearly to both domain experts and cross-functional teams.
Key facts
What you'll do
- Architect and evaluate next-generation multiscale, regional, and global forecasting systems, emphasizing observational data quality and geospatial alignment to model grids.
- Execute cycling data assimilation and forecast loops end to end, including lateral boundary conditions, sea surface temperatures, soil states, and spin-up at convection-permitting resolution over Indian sub-regions.
- Implement rigorous forecast verification frameworks, covering deterministic metrics like RMSE and spectral scores, as well as probabilistic metrics such as CRPS and Brier Score.
- Curate and process ML-ready global and regional weather datasets at large scale, with explicit focus on data-sparse regions and long historical time horizons.
- Procure, harmonize, and validate high-volume, multi-source weather data, including satellite, radar, reanalysis, and station observations.
- Optimize models for renewable-sector applications, specifically targeting solar global horizontal irradiance and wind speed at approximately 100 meters.
- Support the training and integration of AI-based weather prediction models, including learned parameterizations and emulator development.
- Diagnose systematic model biases and tune parameterizations through iterative experimentation and verification against observations.
- Collaborate closely with machine learning and software engineering teams to deploy robust, scalable workflows.
- Maintain reproducibility and production-grade standards across pipelines, leveraging modern data engineering practices.
- Contribute to the design of geospatial processing workflows, including reprojection, masking, and grid transformations.
- Lead the development of benchmarking protocols for nowcasting and extreme event prediction.
- Document methodologies and results to ensure clarity for both technical and non-technical stakeholders.
- Represent Pravah in external collaborations and internal knowledge-sharing initiatives.
Requirements
- Hold a master's or PhD in geophysical sciences, physics, applied mathematics, computer science, statistics, or a related field.
- Accept a bachelor's degree with three or more years of relevant research or operational experience in lieu of advanced degrees.
- Demonstrate deep expertise in numerical weather prediction through operational work, model contributions, research projects, publications, or technical reports.
- Show hands-on experience configuring and running limited-area or mesoscale models such as WRF, MPAS, or comparable systems end to end.
- Include dynamical cores, physics parameterizations, and boundary-layer or convection schemes in your operational workflows.
- Have direct experience running convection-resolving simulations at spatial resolutions approaching one kilometer.
- Possess familiarity with existing operational forecasting models such as IFS, GFS, or BharatFS.
- Have a track record of contributing to or maintaining model code or holding responsibility in an operational or quasi-operational forecasting pipeline.
- Work with terabyte-scale, high-dimensional observational and modeling datasets, including reanalysis, satellite, radar, weather station, and sounding data.
- Apply geospatial data processing skills, including grid transformations, reprojection, and mask generation.
- Demonstrate fluency with the modern geoscience Python stack, specifically xarray, dask, zarr, and netCDF.
- Build reproducible, production-grade pipelines and maintain codebases that support scientific workflows.
- Communicate complex technical concepts effectively to domain experts and cross-disciplinary collaborators.
- Work reliably within distributed teams across India and the United States.
Nice to have
- Bring prior project experience specific to Indian geography and regional climate modeling.
- Engage with coupled earth-system models and understand their limitations for weather-scale prediction.
- Apply ensemble and probabilistic forecasting methods, including subseasonal-to-seasonal prediction.
- Contribute using tools and workflows from operational centers such as IMD, NCMRWF, ECMWF, or NOAA.
- Leverage AI and machine learning techniques in weather prediction, including hybrid physics-ML frameworks.
Practical notes
The team is distributed across India and the United States, so expect a few hours of evening overlap with US Pacific Time on most workdays.