factaday
A small app made using FastAPI to generate and display one random fact, and previous ones, every day using Anthropic's Claude. Only one fact is generated per day and it's triggered by the first user who enters the site that day. Every fact is saved and fetched from a SQLite database and the application is hosted on my personal VPS using Docker and Docker Compose, utilizing Caddy as a reverse proxy to serve the app on my own domain.
source code ->
nightingale
An ongoing research project on cardiovascular health based in Germany. I was part of the Data Science & AI team gathering data on pm2.5 and pm10 particle density in the air and highlightning fundamental links between it and health issues. I used Python to build webscrapers to scrape data from a variety of sources, including community sensors and satellite data. Using Pandas & Matplotlib to understand and refine the data I was working on and creating understandable graphs.
source code ->
skai
A collaborative project with Katrineholm's municipality, Sweden, to both modernize and digitilize their workflows using common practices such as OCR document readers and educations on the usecases of AI. Our most in-depth project was skai, a school system / learning platform with integrated AI, both helping teacher's with assignment creation and correction, and with students regarding common questions about their assignment. We used a model from OpenAI, and using both real and artificial assignment data we finetuned the model to limit it's creative abilities and to make it help the students in a more pedagogical way. This project got national news coverage in Sweden.