Dustin Turner

Search the log and builds

Search posts, builds, and glossary terms.

Operations × Robotics × ML·Est. 2007·Rev B

I run operations alongside the robots. Now I’m learning to build them.

I lead operations inside a heavily automated fulfillment facility, where I watch machine learning meet the physical world every shift. I’m working my way into the engineering behind it — the maths, the models, the robotics stack — and publishing every step of it here.

Monochrome engraving of an industrial robotic arm in a fulfillment warehouse, packages mid-air dissolving into chaotic sketch lines
Fig. 01The demo, dissolving21:9
The arc

Eighteen years from the first floor to the deployment seat. Now the other direction.

01

2007 · First floor

A production engineering internship put me on a floor full of industrial automation for the first time.

02

Today · The deployment seat

Leading operations inside a heavily automated fulfillment facility. The machines run around my team every shift.

03

Now · Crossing over

Linear algebra → probability → ML → the robotics stack. Two public checklists, 2,000+ topics, and an explainer published for every box I check.

The Tuesday Night Test

Every robot passes the demo. The test is Tuesday night.

Vendors sell the demo. Operators live the deployment. Everyone can see the gap between those two now, including the vendors. What I watch, every shift, is what closes it. I call it the Tuesday Night Test: 11pm on a Tuesday, a crumpled mailer, a new hire, a jam upstream, and nobody from the vendor in the building. The machines that survive that night have earned their place on the floor.

The standard
“Machines earn their place on a floor one uneventful Tuesday at a time.”
What to expect

Four things I write about, all from the same seat.

01

Operator's take on robotics news

What the announcement says vs. what the floor will say.

02

Plain-English teardowns

How these machines actually work — the cameras, the grippers, the ML behind a single pick decision.

03

The Tuesday Night Test

What it takes for a machine to earn its place on a live floor: exceptions, night shifts, trust, and the boring work that closes the gap.

04

The learning log

Working through the maths and the robotics stack in public — every topic tracked, every checked box backed by a published explainer.

The learning log

Two checklists, 2,101 topics. You can see exactly where I am.

I wrote the whole map before I started, so the denominator can’t move to flatter the numerator. A box only gets checked when something ships alongside it — so far that’s 16interactive explainers, written for the person I was three months ago. Everything I haven’t touched stays visible, which is nearly all of it.

Topics mapped
2,101
Boxes checked
17
Explainers shipped
16
What I’m chasing

Robotics keeps calling the move from a lab capability to a working deployment its hardest unsolved problem. I already live on one side of that translation. I’m learning the other side in public, and this is the record of it.

Log · Recent writing

From the floor up

Jul 2026The Tuesday Night Test11pm on a Tuesday, a crumpled mailer, a new hire, a jam upstream, and nobody from the vendor in the building. The standard a machine has to meet to earn its place, written from the buyer's side of the PO.drafting
Jun 2026The gap between the demo and the deploymentThe starting point. Why the space between the demo video and the production floor is where I keep looking.shipped
Jun 2026UPS just bet $120M on robots that unload trucksThe spec sheet says 400–1,500 cases per hour. That 4x range is doing a lot of work.drafting
Jun 2026The robot didn't fail. The input did.Why pilots succeed and deployments struggle: demos run on clean inputs, production runs on reality.drafting
The Log · by email

Watch an operator learn the engineering.

1–2 entries a month: a new explainer as I work through the maths and the robotics stack, plus what the floor teaches me that the textbook doesn’t. No noise, no announcements.