Notes on building AI
I build AI systems you can actually explain.
AI engineering, explainability, and honest data analysis. Tutorials, working demos, and what I learn along the way.
-
Why I build explainability in first, not last
A model you cannot explain is a model you cannot defend. Here is how I bake SHAP and LIME into the workflow from day one.
-
Building a RAG chatbot that admits what it does not know
Retrieval is the easy part. The hard part is getting the model to say "not in the sources" instead of inventing an answer.