Engineer

I build systems that have to work.

Physical products, manufacturing, software, AI, and field delivery are different materials around the same work: find the constraint, acquire what the problem needs, build the system, and stay with it until the result is real. Below, the record is arranged by the physics it governs rather than the year it happened.

The constraint field

Matter & HeatMotion & FaultData & AILight, Sound & Perception
99 nodes · hover to identify · click to open

Matter & Heat

High-Stakes Physical Architecture

Motion & Fault

Kinetic Diagnostics & Combustion Logic

Data & AI

Systemic Sovereignty & Code Logic

Light, Sound & Perception

Sensory Architecture & Waveform Physics

The Intersections

The domains are not adjacent specialisms. They overlap, and the overlaps are where the argument lives.

A The Diagnostic Method

The logic that isolates a computer-engine feedback fault is the logic that isolates a thermal failure in a workstation. The machine has knowable properties and discoverable rules.

Matter & Heat ∩ Motion & Fault

B The Zero-Tolerance Threshold

Environments that do not forgive mistakes: cardiac catheters, high-speed kinetic systems, and optics projecting directly onto a retina. Failure here is immediate and physical.

Matter & Heat ∩ Motion & Fault ∩ Light, Sound & Perception

C The Persistent Commit

Hardware rigor translated to digital infrastructure. A manufactured object survives its tooling; an agent's context must survive its container. The container burns, the commit survives.

Matter & Heat ∩ Data & AI

D Total System Parallelism

A consultancy, a repair shop, a race season, and a product schedule running at once. The parallel processing was never a strategy. It was constitutional.

Matter & Heat ∩ Motion & Fault ∩ Data & AI ∩ Light, Sound & Perception

The Long Version

The formal education arrived by correspondence, at the end of Highway 101 on the British Columbia coast, in service of a shellfish cooperative. He burned through it as fast as he could. The real curriculum was outboard motors and tugboat engines and bilge pumps and cable suspension winches, and the particular patience required by living systems that do not care about your schedule. The ocean is an unforgiving, consistent, and honest teacher about consequences, in a way that classrooms rarely are.

At twelve he wanted a motorcycle. He was told he could have one when he could pay for it. Three months later he had felled, bucked, split, and loaded nineteen cords of wood at a hundred dollars a cord, which covered the ferry, the bike, tires, tune-up parts, and a filter. It was never a story about a motorcycle. It was the operating system establishing its core parameters.

When the cooperative failed he moved south with nowhere particular to go, worked a service station, and enrolled at De Anza College - twenty-two units a quarter, an Associate's Degree in Automotive Engine Performance, the Outstanding Academic Achievement Certificate, an Elks National Foundation Vocational Grant. When the schedule required it, he slept in the library. The library was warm and had books.

The first job in technology came in 1985, building a cardiac ablation system - treating irregular heartbeat through minimally invasive catheters. He started as an electronics technician doing assembly, rework, and component-level troubleshooting, and was promoted to Electronics Production Supervisor at nineteen. The work was small enough to hold in your hands and precise enough to kill a patient if you got it wrong. Simultaneously he managed an automotive shop doing computer-engine feedback diagnostics. The same epistemology applied to combustion that was being applied to catheters. He was not building a career. He was developing a method.

In 1989 he joined Silicon Graphics, moving from technician to junior mechanical designer inside a year, learning Pro/ENGINEER and Flotherm and building thermal and mechanical solutions for workstations that consumed power the way race engines consume fuel. Five products from concept to production - the Personal Iris, the Indigo, the Indigo II, the Indy - machines a generation of engineers and scientists and artists used to do their best work.

He founded Mechanistic in 1993. The name was deliberate: the world as systems, as mechanisms, as structures with knowable properties operating according to discoverable rules. Over the following decade he took twenty-two commercial products from start to finish - inline speed skates, a logic-window toaster, a telephone, the Microsoft Xbox, a handheld, an MP3 player, a security appliance, a dental x-ray system. Products people held and used and sometimes loved without knowing his name. That is the nature of the work. The maker disappears into the made thing.

He also owned and operated a repair shop specialising in late-model driveability and racing setups, and raced SCCA SoloII competitively. A consultancy, a shop, a race season, and a product schedule, all running at once. The parallel processing was never a strategy. It was constitutional.

In 2003 he moved to professional audio - the physical architecture of the tools that make the tools: control surfaces and interfaces used to record a great deal of the music of the last thirty years. He was also administrator of the CAD dataservers, already wrestling with persistent, auditable, durable professional data two decades before he had language for the problem.

From 2008 he was the sole mechanical engineer for six generations of high-end audiovisual servers and players, for people who care about the difference between experiencing something and merely consuming it. Around 2010 he discovered photography - landscape, macro, portrait, studio, and a portable photobooth he engineered himself. The same eye that tracked tolerance stacks and thermal budgets, turned toward light and time. He was riding roughly ten thousand miles a year.

Then a wearable video headset projecting images directly onto the retina through beam-splitting optics - engineering the precise boundary between the physical world and human perception. Then a connected light switch and its regulatory compliance program. Then automated foodservice makelines: robots making burritos to the tolerance precision of aerospace assembly, because a burrito is a constraint problem if you care about consistency at scale, and the physical world does not forgive.

He returned to Mechanistic in 2022, four decades of constraint environments having accumulated into something past technical expertise - a forensic epistemology applied across engines, catheters, workstations, optics, food robots, musical instruments, bicycles, photographs, and bodies. He had also noticed that the automated systems now ranking people like him had been built by software people who had never held a thermal resistance budget or managed an injection molding tolerance stack. Their ontologies had no vocabulary for the work. Not failing a screen. Never screened at all. He called it Round Zero, and built an open standard for agentic professional identity to end it.

Around the same time he answered a different question: how do you give an AI agent durable memory that survives sessions, restarts, and complete environment teardowns? The answer was git - not as a workaround but as the natural substrate, because every property required was already there. Agents boot cold, read an issue, execute one task, post a comment, and die.

The container burns. The commit survives.