Selected work
Real systems. Real constraints. Decisions that matter beyond the week they are made.
Infrastructure and cloud work
Designing, stabilizing, and modernizing systems that need to hold up.
Data and database work
Schema design, migrations, cleanup, and performance repair.
Operational clarity
Making systems easier to observe, reason about, and respond to.
Technical judgment
Helping teams make decisions when the answer is not obvious and the consequences linger.
Some problems need research. Some need implementation. Some need a better frame first. Often, they need all three.
Background
13 years building and operating production systems in regulated healthcare — blood center software, clinical data platforms, and infrastructure where downtime had direct patient impact. PostgreSQL at scale. Blue/green deployments on RDS. ETL pipelines processing millions of clinical records. Systems that ran for years without the people who built them.
The practice comes from that work: understanding what holds, what breaks, and what gets expensive when nobody is watching.
Built alongside the practice
We also build products of our own. That keeps the practice honest. The work gets tested in real systems, with real tradeoffs.
What exists. What is missing. What should endure.
What we're building
Collectors lose track of what they own, can't find what they're missing, and have no durable record of either.
We're building three connected products around that problem. CollectionChecker is live now.
We are starting deliberately. The current wedge is Micro Machines: a finite world with enough depth, ambiguity, and collector seriousness to make the model worth proving properly.
CollectionChecker
Start here →Know what you have. See what you're missing.
A collector-facing registry for what exists, what you own, and what is still missing. Structured catalogs, ownership tracking, reports, and cross-house search.
w3f1nd
In developmentFind the missing piece.
A sourcing layer for the gaps in your collection. Wants, briefs, curated seller flow, and deal logic built to follow trust instead of outrun it.
Steleology
In developmentPreserve what should endure.
Provenance, condition, continuity, and transfer-ready records. Built to preserve what matters, not just store files.
Together they describe a single system: what exists, what is missing, and what should endure.
The current surface is intentionally narrow. Get one world right, make it useful, and let behavior challenge the form before widening.
Dealing with something like this?
Contact