Data & AI engineer · Seattle
I build AI & data systems that hold up in production.
Field notes, hands-on tutorials, and real case studies from shipping production systems: the agents, the pipelines underneath them, and the parts that never make the demo.
What today's AI actually costs
The bill moved left, from running software to experimenting with it, steering it, and evaluating it. A field report on where the money really goes, and why so much of it never ships.
Read it →work
Case studies from production work.
All work →A case study from four months building a production AI system inside a Microsoft-native enterprise.
One calibrated model behind every surface, and a chat agent whose first rule is to never invent a number.
A hearing-aware audio platform that measures your hearing with and without earplugs, then filters your music to match, built on a Redis-free Postgres job queue.
A medallion pipeline on AWS (Bronze, Silver, Gold) that cut query latency 16× and data scanned by more than 200,000×, built as a graduate capstone with the trade-offs made explicit.
A ten-year analysis of Seattle crisis contacts that found CIT-trained officers use more force, not less, and a careful account of what that does and does not mean.
Recent essays
All writing →Constitutional AI without the philosophy
A working LLM-judge loop with a bounded retry budget and a visible audit trail when it fails to converge.
When to abandon the low-code path
Why I rebuilt my procurement agent from scratch in week three of a three-month internship.
Anti-hallucination through tool grounding
A small experiment in forcing every claim from an LLM agent through a deterministic database query.
Notes on AI-assisted development
Patterns that have held up across six months of work with Claude Code.
AI engineering
Shipping agents and LLM systems that hold up outside the demo: grounding, evaluation, and the failure modes nobody screenshots.
6 pieces →DataData engineering
The pipelines and infrastructure underneath the models, the unglamorous plumbing that decides whether anything above it can be trusted.
New beat →EngineeringSoftware & craft
How the work actually gets built: AI-assisted development, hard rebuild calls, and the patterns that survive contact with production.
5 pieces →
Adopting AI? I help teams get it into production without the mess.
I advise teams and companies on where AI fits, what to build versus buy, and how to get something into production that survives contact with real users.
- Where AI fits, and where it doesn’t
- Build vs. buy, with the trade-offs made explicit
- From prototype to production that survives real users