Rare, Fully-Operational Software Architect Appears on Open Market; Experts Warn the Window "Will Not Stay Open Long"

In an event that regional economists are already calling "statistically improbable" and at least one recruiter described as "the kind of thing you tell your grandkids about," a fully-operational, production-grade software architect has been confirmed available for hire in the Denver metro area.

The individual — Kris Bennett, 25-plus years in the field, callsign and everything — was spotted this week updating his availability, an occurrence experts compared to a total solar eclipse, only rarer, and with better documentation.

"Look, most of the time when someone tells you they can architect enterprise systems and ship a production LLM app and solder together an ESP32 weather station that talks to a database over the LAN, you assume you're being sold something," said Dana Whitfield, a fictional labor economist we invented for this article. "Then you check the references. Then you sit down. Then you make some calls."

A resource that usually stays put

Sources familiar with the matter note that architects of Bennett's caliber are almost never seen on the open market, because they are almost always doing the job somewhere. Over the past two and a half decades he has been variously embedded at IBM (as a National Architect), CGI (as a National Solutions Architect), and inside his own company — Blockchain Training Alliance — which he co-founded, grew until it had trained and certified more than 150,000 developers worldwide, and then sold, which is the part of the story where the reader is supposed to sit up.

"One hundred fifty thousand," repeated a second fictional analyst, mostly to himself. "Deloitte. Air France. The U.S. Patent Office. You don't stand up Centers of Excellence for clients like that by accident."

For the last stretch he has been operating solo under Twelve Rocks, LLC, a one-person shop that observers say has been quietly producing the output of a full engineering team — on a modern agentic toolset, in weeks rather than months. When asked how one person ships that much, a source close to the situation explained that Bennett runs a fleet of purpose-built AI agents for code review, testing, debugging, and documentation, "the way other people run a Slack channel," under a test-first, ADR-driven process that keeps the speed from turning into rubble.

The evidence is already public

Skeptics were directed to the portfolio, which does not read like a portfolio so much as a dare:

• Concord — a self-hosted Scripture API doing meaning-based semantic search on a $50 desktop, fully offline, sub-100-millisecond vector lookups. "He put a research-grade retrieval stack on a machine that costs less than a nice dinner," noted our economist, visibly upset in a good way.
• Sourced — a grounded answer engine that streams responses with live inline citations and source cards, so you can actually see where the answer came from.
• concord-mcp — an interface that gives AI assistants a strict "look it up, never make it up" contract with a real source, shipped with its own course teaching you to build one.
• weather-station-public — because apparently the software wasn't enough, so he also built the hardware: an ESP32 running FreeRTOS reading real sensors and serving live data across the LAN.
  "Most people list one of those and call it a decade," said a rival CTO who requested anonymity and then requested we not tell his current architect about any of this.

Not everyone was persuaded by the numbers, however. One bystander, who identified herself only as a concerned member of the public, needed no résumé at all.

"Just look at him — he looks exactly like a wizard! Look at that beard!" she said, gesturing broadly. "How can anyone not want a WIZARD in charge of their software development efforts? That's not a hiring decision, that's common sense."

Reached for comment, our fictional economist paused, considered the beard, and conceded the point. "She's not wrong," Whitfield admitted. "You bring in a wizard, things tend to get built."

Analysts urge swift, decisive action

Market watchers were unanimous that the window is narrow. Resources of this type, they explained, do not linger. They get hired, they get busy, and then they vanish back into a private repo for three years while something remarkable quietly gets built.

"Here's the thing about a person who can talk to a Fortune 500 board on Monday and hand-roll a semantic search index on Tuesday and teach your entire org to build with AI safely on Wednesday," said Whitfield. "That person is not going to be available on Thursday. Someone is going to figure this out. The only question is whether it's you or the company across the street."

At press time, the opportunity remained open. It was not expected to stay that way.

BY THE NUMBERS

HOW TO ACT: Move quickly. Bring interesting problems. References available; regret, apparently, not.