Notes on AI, ML & Data Engineering

Practical writing from engagements — what worked, what didn't, and why.

A gray conveyor running between glass frames at night

RAG6 min read

What Makes a RAG System Actually Reliable

Retrieval-augmented generation reduces hallucination in theory. In practice, reliability comes down to a handful of unglamorous engineering decisions.

Abstract streaks of light on a dark background

Data Engineering7 min read

The Data Engineering Work Nobody Wants to Fund Before an AI Project

Every AI roadmap eventually runs into the same wall: the data isn't ready. Here's how to budget for that honestly instead of discovering it mid-project.

Blue and green light shining in the dark

AI Strategy4 min read

When Not to Use an LLM

Generative AI is a good default answer to fewer problems than the current hype suggests. A practical framework for when a simpler system wins.

Abstract plexus of connected blue data nodes

Machine Learning5 min read

MLOps Is What Happens After the First Deployment

Shipping a model is the easy part. Keeping it accurate, monitored, and explainable six months later is where most ML initiatives quietly fail.