About

Ithaca is a public learning practice.

This site is a home for technical writing on LLM systems, agents, evaluation, and the engineering taste needed to build useful AI products.

The name Ithaca is a reminder to keep a destination in mind while doing the day-to-day work: read carefully, build small artifacts, test assumptions, write clearly, and improve in public.

The long-term goal is to develop technical judgment through output. Posts here will favor concrete implementation notes, paper-to-code walkthroughs, evaluation writeups, system design sketches, and honest failure analysis.

Writing themes

  • LLM application architecture and reliability
  • Agent loops, tools, memory, and workflow design
  • RAG systems, context engineering, and retrieval quality
  • Evaluation harnesses and observability for AI behavior
  • Research notes that turn papers into runnable systems