LLMs & language experiments
Turning public news into a small analytics project.
I like experimenting with language models, data pipelines, and public signals. News sentiment is a fun way to combine pipeline work with curiosity about how the world is being framed.
Pipeline flow
NOS articles
↓
Airflow
↓
Ollama classification
↓
PostgreSQL
↓
This portfolio
Today overall
NEGATIVE
Score: -0.4
48 articles
7 day average
NEGATIVE
Score: -0.4
254 articles
30 day average
NEGATIVE
Score: -0.4
1063 articles
Today by category
-
BINNENLAND 0.0
-
BUITENLAND -0.9
-
ECONOMIE -0.4
-
POLITIEK -0.6
-
SPORT 0.2
-
VOORPAGINA -1.0
7 day average
-
BINNENLAND -0.4
-
BUITENLAND -0.7
-
ECONOMIE -0.6
-
POLITIEK -0.3
-
SPORT 0.1
-
VOORPAGINA -0.3
30 day average
-
BINNENLAND -0.5
-
BUITENLAND -0.7
-
ECONOMIE -0.5
-
POLITIEK -0.4
-
SPORT 0.2
-
VOORPAGINA -0.4
Why I enjoy this
It sits right between technical infrastructure and everyday curiosity. I get to work with real data, build the ingestion flow, and then look at the output in a way that is still accessible and human.