cd ~ / blog
~ /blog
Notes on what I’m building and learning.
- Retrieval quality without retrainingMost RAG quality problems are retrieval problems. Reranking, chunking, and eval harnesses beat fine-tuning for far less cost.
- Multi-agent orchestration with LangGraphA practical pattern for LLM-as-router plus specialist agents — decorator-based registration, typed state, and where it breaks.
- Context engineering for production LLMsPrompts are easy; contexts are hard. What to put in the window, what to leave out, and how to budget tokens under real latency limits.