Blog
Thoughts on AI, Engineering, and Building.
Notes on AI systems, retrieval, developer tooling, and the engineering tradeoffs behind useful products.
5 min read
Building Agent Workflows That Stay Understandable
Practical notes on keeping AI agent systems observable, testable, and useful as they grow beyond demos.
4 min read
Measuring Retrieval Quality Before Adding More Models
Retrieval systems improve faster when evaluation starts with the corpus, queries, and failure modes.
6 min read
Designing AI Engineering Tools for Repeated Use
AI tooling should optimize for inspection, comparison, and iteration instead of one-off magic moments.