Machine Learning & AI Engineer with a PhD in Computer Vision for Earth Observations (UCPH, exchange at ETH Zurich), focusing on computer vision, deep learning, and geospatial data for environmental monitoring. With 3.5 years of academic research and 3+ years of professional experience across Europe, the USA, and Asia — now happily exploring AI models and toolings in an embedded software company.
Passionate about deriving insights from big data to solve real-life problems. I also enjoy sharing knowledge through presentations, teaching, blogging, and publishing.
Mastered JavaScript in a week back in 2014 by reading a 687-page guide. Aspiring standup comedian and full-time meme connoisseur.
Python (+ years), JavaScript, SQL, Bash
PyTorch, Scikit-learn, Hugging Face, Computer Vision, Deep Learning
GDAL, Rasterio, GeoPandas, Google Earth Engine, QGIS, ArcGIS
Azure, AWS, Docker, Git, CI/CD, GitHub Actions
PostgreSQL/PostGIS, Pandas, NumPy, xarray, NetCDF, PySpark, Dask, Airflow
Plotly, Matplotlib, Seaborn · English (near-native), Chinese (native), Danish (beginner)
Currently building danskprep — a fun side project to help with Danish language exam preparation, built with TypeScript and deployed on Vercel.
An app for Danish exam preparation.
Dark terminal-themed personal portfolio — particle canvas, neko pet, cat/dog vote (Supabase), auto-synced GitHub & Google Scholar data, multi-language (EN/ZH/DA). Pure HTML/CSS/JS, no frameworks.
Browser-based pixel art editor with AI generation (Gemini) and MCP server for AI agents. React + TypeScript + Vite.
No description
Run Vision(-language) models for earth observations at fingertips
Browser-based pixel art editor with AI generation (Gemini) and MCP server for AI agents. React + TypeScript + Vite.
An app for Danish exam preparation.
Cross-resolution segmentation of individual dead trees from aerial images
End-to-end framework for a customized UNet-based pipeline for large-scale individual dead tree mapping
A list of companies focusing on geospatial intelligence, GIS, RS, Climate risks, and more
How OpenAI shipped code with zero manually-written code — and what it means for how we design environments for AI agents.
5-month internship at Esri headquarters in California
Learning by doing — no prior knowledge required!
Save yourself from endless errors while installing packages and running old scripts.
A guide to keep Docker images lean, secure, and reproducible.
Very first step to get your deep learning model up and running on GPU.
Better managing raster timeseries using GeoServer
Make your ETL pipelines robust to changes and scalable for big data processing.
Leading companies harnessing geospatial intelligence and AI for decision-making across industries.
Publications and projects from my academic years — currently happily exploring AI models and tooling in the embedded software industry.
First globally to use PlanetScope for phenological analysis over semi-arid grasslands
read moreTested and documented the functionalities of the TRAILS platform
read moreResearch, GIS analyses, new indicators/layers, and Esri Story Maps demonstrations
read morePython packages for cloud-based preprocessing and time series analysis of Landsat data
read more306 citations · h-index 5 · 19 publications