Personal projects / Oslo, Norway
Senior Data Engineer at DNB. In my own time I build open tools, local AI services and small experiments where software meets hardware. I like technology that people can inspect, test and actually use.
Build · Explore · Share
01 / Personal projects
Things I have built outside work. Most of them are live or have public code.
A Norwegian election quiz that answers each statement with what the parties actually voted for, proposed or passed, not what their programmes promise. 57 statements, 589 evidence files, and a script that checks every quote and every vote against its source.
Go · YAML · Python
Dashboards, reports and slides as one HTML file you can drop in SharePoint or OneDrive. No licence, no server, no install. 4 million rows fit in an exact 0.81 MB data cube. A colleague's idea, I build it.
Python · JavaScript · HTML
A neon Tetris clone in one HTML file and one JS file. No libraries, no assets, no server logic. Made for a party, still live.
JavaScript · Canvas · Go
A written proposal for a Nordic digital commons. Shared infrastructure, tools and AI, owned in common and run at cost. Nothing is founded yet. This is the design work, in plain Norwegian, with drawn figures.
HTML · CSS · Go
My flat runs on a git repo. Five small machines pull their own config from it: Home Assistant, a local voice assistant, a GPU box serving open-weight models, and the server behind this page. Seven public sites, each one Go binary in an empty container.
Docker · Home Assistant · Tailscale · Cloudflare Tunnel · vLLM · Go
69 hours, three tasks, one team. Our Astar Island solver sat at #8 of 214 teams during the competition. Also grocery object detection with YOLO and ONNX, and an LLM agent that does accounting.
Python · YOLO · ONNX · LLM agents
Turned a wall of SQL/VQL metadata into a graph you can actually look at. Filed a pandas bug along the way.
Python · SQL · PostgreSQL · Denodo VQL
Same idea, in an app you can double-click.
Tauri · React · Rust · Python
A handful of Norwegian summarization/translation datasets and a translation model, still sitting on Hugging Face from the thesis days.
Hugging Face · Transformers · PyTorch
02 / How I work
The interesting part is usually between the pieces. The data, the model, the API, the hardware and the person who has to use it.
If something matters, I want to know how it works and how it can fail. Especially when AI is involved.
A tool is worth more when others can inspect it, use it and build on it.
03 / Background
Senior Data Engineer, Insight Factory
Jun 2026 – Present
Data Engineer, Group Data Insight
May 2023 – Present
Data Engineer, ESG Data Hub
Jun 2023 – Aug 2023
Full Stack Developer (consultant)
Sep 2022 – Jan 2024 · Ås, Norway · Hybrid
Built a system using electric vehicles as batteries to stabilize grid current draw.
Teaching Assistant
Aug 2021 – Dec 2021 · Ås, Norway
TA for data structures and algorithms.
Machine Learning Engineer (consultant)
Jun 2022 – Aug 2022
NLP on textual appraisal reports, on a team of 7.
AI & ML Engineering Intern (consultant)
Jun 2020 – Jan 2022
Full-stack ML: from data warehouse to an application ordering predictions from deployed models, across 30+ teams.
MEng (Sivilingeniør), Data Science
Aug 2018 – May 2023
Business Analytics specialization. Grade A.
MEng exchange, Computer & Information Sciences
Jan 2022 – Jun 2022
École Supérieure d'Ingénieurs Léonard de Vinci, Paris.
04 / Contact
An idea, a problem, a prototype, or just something we can learn from each other.
