Projects & Research

Open Source Projects

pyaerocom

Python Tools for Climate and Air Quality Model Evaluation

Contributed to this comprehensive Python package for processing, analyzing, and visualizing aerosol and air quality model data. Serves as the backend for the Aeroval evaluation portal and is used operationally by the Copernicus Atmosphere Monitoring Service (CAMS).

• Tech Stack: Python, xarray, pandas, cartopy
• Contributions: Data fusion algorithms, model evaluation metrics, uncertainty quantification
• Impact: Powers operational air quality evaluation portal, used by European forecasting services

DeepTreeMRA

Scalable Spatial Modeling for Massive Datasets

High-performance implementation of the Multi-Resolution Approximation (MRA) for spatial statistical modeling. Designed to handle hundreds-of-millions of observations using deep tree structures and parallel computing.

• Tech Stack: MATLAB, MPI, High-Performance Computing
• Key Features: 75% computation time reduction, scalable to 100M+ observations
• Applications: Climate data analysis, environmental modeling, large-scale spatial prediction

optimparallel

Parallel Optimization for Python

Python package providing a parallel computing interface to the L-BFGS-B optimizer, enabling efficient optimization of complex objective functions.

• Tech Stack: Python, NumPy, parallel computing
• Use Cases: Parameter estimation, model fitting, optimization problems

Professional Work

Statkraft (Current)

Building the ETL pipeline underpinning Statkraft’s flagship long-term energy market model — internally described as the most significant change to the platform in fifteen years. Focus areas include:

• Kubernetes health-check redesign and observability infrastructure that cut production incidents from multiple per week to near zero
• The team’s first Kubernetes-based test environment, built with Kustomize
• Upstreaming a patch to the open-source FIMEX library to unblock an OS migration

Norwegian Meteorological Institute

Led development of machine learning models for air quality forecasting and contributed to the Copernicus Atmosphere Monitoring Service (CAMS):

• Machine learning models for air quality forecasting and uncertainty quantification methods for chemical transport models
• Pollution deposition modeling for Norway and air pollution modeling for the Netherlands using ML-based data fusion techniques
• Lead developer of pyaerocom (see Open Source Projects above)

NCAR Research

Created scalable algorithms for massive spatial datasets:

• Parallel computing methods reducing computation by 75%
• Analysis of satellite data (100M+ observations)
• Statistical modeling for climate and environmental data

Technical Interests

• Quantitative Modeling: Time series analysis, forecasting, optimization
• Machine Learning: Deep learning, ensemble methods, statistical learning
• High-Performance Computing: Parallel algorithms, distributed systems
• Data Engineering: Large-scale data pipelines, real-time analytics