Noos - Where Thought, Code, and Craft Converge

A proposal-style exploration of a rule-based, structure-aware prompt compression engine for Tokuin—and a call for expert feedback.

A look at Tokuin, a ground-up token analysis and load-testing CLI for LLM prompts, why it exists, what it offers today, and where it's headed next.

mock is a lightweight API framework that lets you build endpoints using shell scripts—perfect for prototyping, testing, and quick integrations.

Comparing ROUGE-L implementations in Java and Rust: same algorithm, same results, dramatically different performance. What I learned from this fun experiment.

Observations from building AI systems: why prototypes struggle to become production-ready, and what the research shows.