Bart Olsthoorn
Ph.D. in Physics · Product Engineer at Flower · Founder of Glimt
Now
I'm a product engineer and build software end-to-end, from data pipelines on Databricks to the APIs and web interfaces on top. At Flower the domain is distributed energy assets: batteries, solar, EVs and heat pumps. Lately I build a lot of it with AI agents.
On the side I founded Glimt, which builds scientific instruments in Sweden. The first is a scanning tunnelling microscope for atomic-resolution imaging.
Experience & education
- 2026–now
- Glimt Founder · Scientific instruments, starting with a scanning tunnelling microscope · Sweden
- 2023–now
- Flower Product Owner / Engineer · Distributed Assets · Stockholm
- 2018–2023
- KTH Royal Institute of Technology Ph.D. Physics · thesis: Homology and machine learning for materials informatics · 11 peer-reviewed publications
- 2015–2018
- Stockholm University M.Sc. Computational Physics
- 2012–2017
- Freelance ML at Similar.ai (early GANs), Ruby/Elixir at Universal Avenue, Rails at InnerBalloons & Pointer Brand Protection
Stack
- Core
- Python, Go, full-stack web (APIs & interfaces), AWS, Databricks, PostgreSQL
- AI
- Agents and agentic loops in production · PyTorch · Pandas / NumPy / scikit-learn
- Past
- Ruby / Rails, Elixir, Swift / Objective-C
Open source
My repositories on GitHub have 550+ stars between them. A few of them:
- NVDSP iOS/macOS audio DSP library · 418★
- NQS-numpy Neural-network quantum states in NumPy · 35★
- simhash-rs Rust implementation of SimHash (locality-sensitive hashing) · 25★
- gohighs Go bindings for the HiGHS optimizer
Selected publications
- Persistent homology of quantum entanglement Physical Review B, 2023 · Editors' Suggestion
- Band gap prediction for large organic crystal structures with machine learning Advanced Quantum Technologies, 2019
Full list on Google Scholar.