DevOps & Machine Learning Expert

Role Overview

The DevOps & Machine Learning Expert will help build and run ICPAC’s regional flood and drought forecasting services, ensuring that data, models and public-facing products work reliably through regular forecast cycles. Reporting to the DRM Program Head, the role combines geospatial software development, cloud operations, hydroinformatics and applied machine learning to support timely warnings and stronger early-warning capacity across East Africa.

Key Responsibilities

• Automate and maintain operational hydrological forecasting chains, enabling new product releases while keeping services available.
• Develop harmonised hydrological reference and observation datasets, quality-control processes and data-exchange interfaces for national hydrological services.
• Build and operate STAC-based data access and analysis services, with dependable processing pipelines for hazard, exposure and impact models.
• Bring East Africa Flood Watch into public operation, integrate its warnings and products with the East Africa Hazard Watch Portal, and support its development toward parity with Drought Watch.
• Automate geospatial processing, reporting and impact-based bulletin production; assess and govern the use of language models in drafting workflows.
• Deploy calibrated forecast post-processing, multi-model ensemble methods and a Bayesian Network risk model, integrating services with risk-monitoring applications.
• Establish forecast verification and routine skill reporting, document operational model governance, and contribute event-based climate storylines to the drought and flood event catalogue.
• Provide technical support to Member States, national meteorological and hydrological services, and regional partners; maintain records of support requests and deliver technical documentation, training and hand-over materials.
• Maintain reproducible multi-cloud infrastructure using infrastructure as code, workflow schedulers, CI/CD and GitOps; monitor production systems and test disaster-recovery procedures.
• Carry out other related duties as assigned.

Requirements & Qualifications

• A university degree in computer science, geoinformatics, hydroinformatics, computer engineering, data science, software engineering, information technology or a related discipline. A postgraduate degree is an advantage.
• At least four years of relevant experience designing and developing geospatial applications, with a portfolio that includes links to at least two previous projects or GitHub code samples.
• Demonstrated ability to operate cloud services in production under daily operational deadlines and to deploy machine-learning models in operational settings.
• Strong Python skills for scripting and geospatial data processing, alongside sound SQL and PostgreSQL/PostGIS knowledge.
• Experience developing web and geospatial applications using HTML, CSS and JavaScript, and tools such as Node.js, React, Mapbox GL, Leaflet, GeoServer, MapServer and GDAL. REST APIs and microservices experience is expected; Go is an advantage.
• Knowledge of OGC standards, including WMS, WFS, WCS, WPS and Simple Features for SQL, as well as experience handling Earth observation data in multiple formats.
• Practical familiarity with STAC APIs and PySTAC; ecFlow and Prefect; xarray, dask and NumPy; rasterio and geopandas; GRIB2, Zarr, COG, VirtualiZarr, Icechunk and kerchunk; and TimescaleDB.
• Hydroinformatics experience covering operational rainfall-runoff, hydrodynamic or rapid-inundation forecasting; forcing-data preparation; catchment, river-network and terrain data; hydrometric and satellite observations; and hydrological data exchange.
• Experience with Docker, Kubernetes and Helm; GitOps delivery such as ArgoCD; GCP and AWS; Terraform, Coiled and CI/CD; and GPU environments, observability, incident response, cloud security and cost management.
• Knowledge of impact-based forecasting, climate-model workflows, ensemble calibration and downscaling, Bayesian networks, forecast verification, MLOps, and evaluation and safeguards for language-model applications.
• Self-motivated and delivery-focused, with sound problem-solving, teamwork and communication skills, and a commitment to continuous improvement and sharing knowledge.
• Experience supporting national institutions in operational early-warning settings is desirable.

What We Offer / Why Join

This role contributes directly to regional flood and drought monitoring, operational forecasting and early-warning support for IGAD Member States. The position also offers the opportunity to work across geospatial technology, hydrology and machine learning in services used by national and regional partners.