AI
Retrieval, evaluation, and building AI products that behave reliably.
Writing on RAG architecture, LLM evaluation, embedding pipelines, and applied AI system design.
Articles
AI writing.
Building retrieval systems that fail predictably
Chunking, embeddings, hybrid scoring, evals, and guardrails for practical AI application architecture.
Read article →Evaluating LLM outputs before they reach production
Building eval pipelines that catch regressions, measure groundedness, and flag hallucinations at scale.
Tool use and trust boundaries in agentic systems
How to scope agent permissions, validate tool outputs, and design agentic workflows that stay in bounds.