Customer churn modelling
An end-to-end case study covering data quality, feature design, model evaluation, and decisions.
Data science student · Building in public
Making data useful,
one clear decision at a time.
I’m Lucian. I document practical work across analytics, machine learning, data engineering, and applied AI.
Selected work
Projects in development
Analytics pipeline
A reproducible pipeline that turns raw source data into tested, decision-ready models.
RAG evaluation lab
A practical exploration of retrieval quality, grounded answers, citations, latency, and cost.
Field notes
Writing about the work
Why accuracy is rarely the whole answer
A practical note on choosing evaluation metrics from the decision and cost of error.
SQL patterns I use in analytics work
Window functions, retention cohorts, rolling metrics, and session boundaries.
Working principle
“A useful model is not the most complicated one. It is the one that helps someone make a better decision.”