Writing
Essays and posts on technology, AI, engineering leadership, and building better systems.
Essays
Long-form deep dives on important topics
Why RAG is Harder Than It Looks
Retrieval-Augmented Generation seems simple in demos but breaks in a dozen ways in production. Here's why most RAG projects fail and what to do about it.
The Hidden Cost of Technical Debt
Technical debt isn't just about messy code—it's about compounding decisions that slow teams down over time. Here's how to measure and manage it effectively.
Building AI Systems That Don't Embarrass You
Your AI system will have its worst moment in front of your most important user. Here's how to build systems that fail gracefully instead of spectacularly.
Building AI Systems That Users Trust
As AI systems become more prevalent, user trust becomes the limiting factor for adoption. Here's what I've learned about building trustworthy AI products.
The Unglamorous Parts of AI Engineering
Twitter shows AI demos. Production requires data pipelines, permission systems, eval harnesses, and on-call rotations. Here's what AI engineering actually looks like.
Posts
Shorter articles and tactical insights
How to Debug a Failing RAG Pipeline
Your RAG system is returning bad answers. Here's a systematic approach to find out why and fix it.
Prompt Engineering Patterns That Actually Work
After writing thousands of prompts, these are the patterns that consistently improve results. No magic tricks—just engineering principles.
Building Your First LLM Agent with Tool Use
A practical tutorial on building an LLM agent that can use tools. We'll build a simple research agent that can search the web and summarize information.
Five Mistakes I Made Building AI Chatbots
I've shipped chatbots that worked great and chatbots that embarrassed me. Here are the mistakes I've made so you don't have to.
Testing AI Systems Beyond Vibes
Looks good to me is not a testing strategy. Here is how to build a real evaluation framework for AI systems that actually catches problems.
When to Use RAG vs Fine-Tuning
Two approaches to customizing LLMs for your use case. Here's a practical decision framework for choosing between RAG and fine-tuning.
Five Principles for Scaling Engineering Teams
Lessons learned from scaling teams from 5 to 50+ engineers while maintaining productivity and culture.