Magnus Strandgaard

Linux and Cloud Operations Engineer

Netcompany

• Maintained and extended the AWS infrastructure for a 1M+ user application across 11 environments (10 dev replicas + production), defining compute, networking, and data resources as code in Terraform with per-environment workspaces and AWS SSO access.
• Became the team's lead Terraform developer within 3 months of joining. Led the zero-downtime migration of an enterprise solution (1,000+ live resources) to a managed Terraform state, establishing rigorous code review practices and mitigating edge cases for seamless production deployment.
• Developed and maintained automated CI/CD pipelines in Azure DevOps and Jenkins, integrating strict policy checks and testing to ensure high-quality, production-ready code delivery.
• Python development for automation of internal processes, including ingestion and processing of critical data. Engineered a robust Python data-processing pipeline that replaced a legacy service, resolving performance bottlenecks and reducing cloud spend by ~$3,500/month while ensuring secure data handling.

Postdoctoral Researcher

Dept. of Chemistry, University of Copenhagen

• Designed and operated data pipelines across 4 HPC clusters to ingest, curate, and validate 100K+ molecular records from the Cambridge Structural Database, enabling structure-property analysis of transition-metal complexes.
• Built ML workflows combining molecular fingerprints and graph neural networks to predict chemical properties of 3D structures, with accuracy gains over baseline models. Orchestrated distributed GPU training via SLURM and the WandB MLOps tool.
• Developed a novel synthetic accessibility scoring tool in Python with C++ bindings for generative molecular design pipelines.

PhD Researcher

Dept. of Chemistry, University of Copenhagen

• Developed end-to-end in silico molecular discovery pipelines coupling quantum chemistry with generative ML models (genetic algorithms, variational autoencoders) for catalyst design and compound optimization, screening thousands of candidates with RDKit, Pandas, and SQLite.
• Collaborated with the University of Oslo theoretical chemistry group to implement a variational autoencoder (VAE) for molecular inverse design, training PyTorch models on HPC GPUs to generate novel transition-metal complexes.
• Served as HPC cluster administrator for the physical chemistry department (50+ active users), managing Linux environments, SLURM job scheduling, and reproducible computational workflows.
• Created and maintained 4 open-source research repositories on GitHub as lead developer.
• Mentored 60+ undergraduate students as teaching assistant for physical chemistry and mathematics courses.

Research Assistant

Dept. of Energy, Technical University of Denmark

• Implemented a Message Passing Neural Network (MPNN) in PyTorch to predict forces and energies of magnesium battery cathode candidates, enabling molecular dynamics simulations of next-generation energy-storage materials.

PhD, Computational Chemistry

University of Copenhagen

MSc, Physics and Nanotechnology

Technical University of Denmark

BSc, Physics and Nanotechnology

Technical University of Denmark