MARK MYDLAND Book 30 min
MYDLAND — ENGINEERING OS Teams healthy · No fire drills REV 2026.07
mark@pnw ~ %
No fire drills. Great teams don't need heroics.

Products with a pulse. Shipped at world scale.

I'm Mark Mydland, VP/SVP Engineering. I build the engineering system that lets great people do their best work: predictable, agile, durable. Then we point it at products that do real good in the world.

Book 30 minutes The record ↓
The recordMicrosoft · Strella · Dexcom
From $0
$250M/yr
in Azure compute: four people to ~5% of all Azure VMs (Azure Lab Services)
From $300M
$1B+
platform revenue, leading ~225 engineers across 55 regions (Azure IoT)
From $0
$130M/yr
Stelo, the first FDA-cleared OTC glucose monitor (Dexcom, reported FY25)
From an 18-month plan
4 months
to shipped, at Strella (Series A), by fixing the rhythm before the code
Case entriesFour problems · One operating system
Founder & Engineering Leader

Built it from four people into the service creating about 5% of all Azure VMs, roughly $250M a year in Azure compute. We invented the auto-shutdown feature Azure adopted platform-wide, and delivered in about half the time anyone expected.

4 people → ~5% of all Azure VMsAuto-shutdown, adopted Azure-wideShipped in half the expected time
Engineering Leader, Core Services

Led the core services of one of the world's largest IoT platforms: IoT core, Device Provisioning, Digital Twins, Edge runtime, SDKs. About 225 engineers across 55 regions, serving customers like Daimler Trucks, Starbucks, BMW, and Bosch.

~225 engineers · 55 regionsDaimler · Starbucks · BMW · BoschSI + silicon deals: ST Micro, NXP, Renesas
Engineering Leader (Series A)

Walked into a Series A in full chaos and left it thriving: the eighteen-month plan delivered in four, three SaaS products in market, inventory shrink cut about 50% for customers, deployed across three continents. The lever was a weekly demo to the CEO and a hard rule that done means production. Rhythm first, then the code follows.

Chaos → 3 SaaS products in market~50% inventory shrink reduction3 continents
VP of Engineering

Delivered Stelo, the first FDA-cleared over-the-counter glucose monitor and the first CGM with AI, opening a new direct-to-consumer market and taking it from zero to $130M a year and climbing (Dexcom's reported FY25 figure). Led 125+ engineers under medical-device regulation with hard deadlines, holding 99.9% crash-free the whole way.

First FDA-cleared OTC glucose monitorFirst CGM with AI99.9% crash-free · 125+ engineers

"Organizations that value firefighters breed arsonists."

Why the system below exists
The frameworksHow the system actually works

Every leader says they have a system. These are mine, named and drawn, because a system you can't hand to someone else isn't a system yet. Each one was built inside a real team, under real pressure, and each one still runs today.

The Airlock

Fig. 1 · Product–Engineering partnership

One shared board with two airlocks. Neither door opens alone.

BACKLOG product's, outright IN PROGRESS engineering's, outright SHIP production ON DECK door 1: shovel-ready door 2: accepted IN REVIEW door 1: dev pushes door 2: PM signs off ~25% OF THE BOARD RESERVED FOR ENGINEERING HEALTH, WATCHED TOGETHER EVERY GROOMING

The chronic product-versus-engineering fight isn't a personality problem. It's a board-control problem: it lives wherever one side can move work across a hand-off alone. So put an airlock on each hand-off. Work enters only when product's door (right priority, shovel-ready) and engineering's door (clear, right-sized) both open, and nothing ships on a developer's word alone. This is the operating model underneath four people becoming ~5% of all Azure VMs.

The Airbag Rule

Fig. 2 · Recovery by design

It's too late when the crash happens.

COST OF BEING WRONG TIME → THE CRASH (IT'S COMING) HOPE + ROLLBACK THE AIRBAG RULE ↑ small cost up front: the airbag

Everyone says "assume failure." Almost no one funds it. The Airbag Rule: believe the break, pay up front to wire recovery in as forward motion (a flag, a switch, reaching every layer), and release at minimum blast radius. Rollback looks safe until the system is complex; the senior move is to never need one. The tell for whether a team really believes failure is coming is the wallet.

False Green

Fig. 3 · Signal over coverage

A passing signal you can't trust is worse than an admitted gap.

THE WALL OF GREEN ? why is any of this green? "COVERAGE": HIGH · TRUTH: UNKNOWN SIGNAL YOU CAN TRUST PASS: knownPASS: knownFAIL: known PASS: knownFAIL: knownPASS: known every tile has a reason attached COVERAGE: LOWER · TRUTH: EARNED

A flaky test is worth less than no test. When it passes you get confidence you didn't earn, and you stop looking at something that might be broken. So judge every activity by the trustworthy signal it produces, delete the work that manufactures the look of quality, and redeploy the people (not cut them) to work that produces real signal. Reds are honest. Greens have to earn it.

Show It, Don't Say It

Fig. 4 · Progress you can verify

You can't fake a demo.

PROGRESS TIME → 100% "50%… 75%… 82%… 84%…" (never arrives) SHIPPED SHIPPED SHIPPED = DONE

"I'm 82% done" is a story; a working demo in production is a fact. Install a recurring demo with a real audience and an immovable date, make done binary (done means production, not a branch, not staging), and read the team's true velocity off what actually shipped. Fix the rhythm first; the rhythm changes the code.

01 / 04
The four questionsTap one
4.1Do you ship at least every day?+
Deployment frequency, in plain language. A team that ships daily has already solved a hundred problems (build health, test trust, small increments) that a team shipping quarterly is still arguing about. A bad answer here usually means the increments are too big to finish.
4.2How long does it take to ship a thing?+
Lead time for changes. The gap between "the code is written" and "a customer can use it" is where organizations hide their dysfunction. Weeks of lead time means approvals, hand-offs, and fear own the calendar, not engineering.
4.3How often do you break it?+
Change failure rate. Some breakage is the price of moving. The bad answer isn't a number, it's a shrug: a team that doesn't know its failure rate isn't measuring the machine, which usually means it's measuring the heroics instead.
4.4How fast can you un-break it?+
Recovery time, and the one I care about most. This is where the Airbag Rule lives (Fig. 2). A team that recovers in minutes designed for the crash before it happened. A team that recovers in days is betting the product on being right every time. Nobody is right every time.

These map one-to-one to the DORA metrics, stated the way I actually ask them. They land with executives who would glaze over at "change failure rate," and a first conversation built on them gives you a fast, credible read on how mature the machine really is.

The personHow I lead

The system serves the people, not the other way around.

Fire drills feel productive and burn out your best people. So I build the operating system that makes shipping boring: anchor the team to the customer, ship the smallest increments that actually deliver, never let quality slip. Incomplete is fine; wrong is not. What that buys you is a team that trusts each other, tells the truth, and clears a high bar because they want to. Excellence without heroics.

Before any of this: West Point grad, Airborne and Ranger qualified. I learned to work on a Wyoming dairy farm and learned what tired means at a jungle course in Panama. These days I'm in Bremerton, Washington, still chasing the best breakfast in Monterey.

Start the conversationNo form. Just a calendar.

I'm looking for the next thing to build, at full scale. Something I can point to and say it does real good in the world.

If you're standing up a new product line, an internal startup or AI innovation group, or a business with an exit to build toward, and you want someone who can build it and make it ship, let's talk.

Book 30 minutes Resume (PDF) LinkedIn Email