Blog
Roll for Insight: What Dungeon Masters Can Teach Us About RAG
The architecture of retrieval-augmented generation maps surprisingly cleanly onto what a well-prepared Dungeon Master does at the table. Retrieve first, then generate — grounded in the source text.
Gradient Descent and Voice Leading: Parallels Between Machine Learning and Classical Music
The same structural ideas keep appearing in both ML theory and classical composition — optimization, overfitting, long-range coherence. The parallel is precise enough to be useful.
Knowing the Grain: Parallels Between Machine Learning and Woodworking
Inductive bias, distribution shift, and the discipline of understanding your material before acting on it — the same ideas show up in the workshop and in the training loop.
FTE Adjustment Model: Modeling Productivity Ramps During Employee Leave
A four-phase algorithm for forecasting employee capacity during extended absences, with SQL implementation and interactive charts.
From BI Dashboards to ML Pipelines: Changing Career Tracks
Reflections on transitioning from business intelligence to machine learning engineering, and why the jump is smaller than you think.