Chris Murphy
Lead Software Engineer | Applied AI & RAG
I build AI-powered applications and write about what I learn along the way. Currently focused on RAG systems, LLM safety guardrails, and turning real development problems into engineering insights.
What I'm Building
Featured Project
AeroStream Manufacturing Bot
A Retrieval-Augmented Generation (RAG) system built for manufacturing technicians and support engineers. It provides instant, cited answers to technical questions about drone maintenance, assembly procedures, and troubleshooting — pulling from SOPs, technical notes, and tribal knowledge.
More projects coming soon
Writing
Articles & Insights
Why Your RAG System Can't Connect the Dots: Embedding Models Matter
RAG retrieval is a three-legged stool: the embedding model, the chunking strategy, and the retrieval process. Weaken one, and the whole thing topples.
Read articleToken Economics: What I Learned Burning Through My Claude Rate Limit
How a 936-line architecture document taught me that AI development costs belong in your architecture decisions from day one.
Read articleEvaluating RAG Systems: Lessons from Building My Own Eval Pipeline
Why automated eval systems fail when answers are better than expected, and what the RAG Triad framework actually looks like in practice.
Read articleThe Gap Between Knowing and Doing: How AI Fixed My PKM System
Years of reading about PARA didn’t fix my broken knowledge system. A 5-minute AI conversation did.
Read articleAbout
Background
I spent over a decade in enterprise software (SAP) building complex systems for large organizations. Now I apply that engineering discipline to AI — building production-grade RAG systems with safety guardrails, multi-strategy retrieval, and automated evaluation pipelines. I write about the real problems I encounter: token economics, RAG evaluation, and the gap between AI demos and AI systems that actually work.