Ideas leave the labwith their edges intact.
Published elsewhere
Recent articles
AI engineering
I Don’t Engineer Prompts. I Engineer Where the Words Sit.
A while back I noticed something about my own prompts that I couldn’t explain at first. The good ones, the ones that actually produced something I could publish without a full rewrite, weren’t the longest ones....
System design
The AI Doesn’t Forget. It Just Remembers Wrong.
How long conversations quietly mislead the AI and why it matters to everyone.
Agent systems
Agentic AI in 2026: What Actually Ships vs What Gets Canceled
A grounded look at where agent systems create durable value—and where demos collapse under production constraints.
RAG
5 Prompting Patterns That Actually Fix RAG Agent Failures
Five production patterns for grounding retrieval agents, each mapped to a concrete failure mode.
Tutorial
Build Your Own PDF RAG Agent: Python, Streamlit, ChromaDB, Neo4j & Groq
RAG agents answer questions from your documents. Upload a PDF, ask questions, get answers with sources. This tutorial builds one from scratch.
Substack
agents / runtime
Agents fail on authority, not intelligence
The interesting failure is never a wrong answer. It is a correct answer applied somewhere nobody agreed to.
retrieval / public sector
Retrieval is an editing problem
Ranking, chunk boundaries, and knowing when to say the data does not cover this.
enterprise / public sector
Government software is judged on the worst day
Five years building systems that cannot be down when a resident needs them.
tech / writing
Read my latest on Medium
Writing about AI, software architecture, and the intersection of technology and creativity.
tech / coding
Check out my DEV Community posts
Technical deep dives, tutorials, and open source contributions.
tech / coding
My Hashnode Articles
Thoughts on web development, system design, and building for scale.