Writing

Notes on machine learning, computer vision, and interpretability.

Should Developers Care about Interpretability?

Why mechanistic interpretability research matters beyond alignment — and what it means for people building with models today.

2 min read · November 06, 2025

2025 · interpretability ml · ml
LLM-Powered Sorting with TrueSkill

Using pairwise LLM comparisons and the TrueSkill ranking system to sort items that resist direct scoring.

2 min read · February 11, 2025

2025 · llm ranking · ml