For years, my career progression looked strange if you tried to read it as a ladder. I loved the adventure of it: every new assignment brought an entirely new set of problems, new fires to put out, and a different part of the machine to understand. I moved through seller onboarding, sourcing, freight forwarding, customs processing and inbounding. Then I worked in Supply Operations with Missing From Amazon. I became deeply familiar with Seller Central, supporting sellers and working in backend T3 Support. Back in the fulfillment sites, I established processes in ISS and ICQA, developed inbound processes and systems, and moved into stow. From there I went into pick, then ship dock, then outbound as a whole. After that came network processes, systems implementations, whistle stops, new site launches, regional high-volume-event mitigations, ROC work, equipment management, investigations, logic systems, software development and operational architecture.
On paper, that might look like a great deal of bouncing around. That is not really what happened. I was following the bottleneck. The pattern was simple: fix system A, and system A becomes capable of producing more. System B is then exposed as the new constraint. Go fix system B. Then C. Then another building. Then a region. The work kept moving because the constraint kept moving. Making one site strong within a region exposed the next opportunity. I did not need a career ladder. The system kept telling me where I was useful next.
Amazon’s principle of helping the bottom ten percent happened to align with the way I worked. I became unusually good at helping sites produce in the top ten percent, which quickly exposed where risk remained for the bottom ten percent. The point was never to make a metric look impressive in isolation. The point was to increase the capability of the system until the next weakness became visible enough to address.
THE FIRST SIDE OF THE MACHINE
Before Amazon, and before the work acquired that name, I had already learned to think in systems because I had needed systems to survive. I came into Amazon from the seller side, but Amazon recruited me; it was not a job I had gone looking for. After the Army, computers were a form of therapy. Seven years of service, much of it in active combat, had left the prospect of returning to ordinary life feeling less like a homecoming than another environment I might not survive.
Combat had taught me to live with the fact that I could die at any moment. Once I survived long enough to move past the first paralyzing terror, I began prioritizing differently. I acknowledged that survival was not guaranteed and that the chance of not surviving was real. From that position, I tried to calculate what I called equitable behavior: what choice would do the most good, given the conditions in front of me? I decided that my best chance of longevity was to remain unreasonable and unpredictable on purpose. I would choose the action nobody could presume, guess or calculate. I believed that unpredictability would buy me enough time to do useful work.
The Army kept sending us children, some of them not yet twenty, who had signed up for weekends with the National Guard and had never held a serious job. I watched them enter situations that produced absolute, paralyzing fear, and I watched some of them become bodies in the morgue trucks that were simply part of the convoy. I expected never to return to civilization. I made choices that crossed lines I believed excluded me from normal human society forever. I did not expect to become somebody who would later have to live with what those choices meant.
When Saddam was captured, six of us remained who had accumulated that kind of long-standing tenure. We could look at one another and know that none of us expected to go home. No one else could look at us and share that particular knowledge. We were more frightened of living as ordinary people in an ordinary world than we were of staying in combat. The end was not a clean release. It was devastating. Standing there with a rifle and realizing that the real combat might be the attempt to become a citizen again was almost harder than adapting to war had been.
The flight home, scattered through the fuselage of a C-130, became more painful the closer we came to the possibility of a hero’s welcome. The weight accumulated. Returning home was supposed to be the reward, yet home represented another environment for which we felt unqualified. Afterward, the unreasonable decisions continued, although the bullets and enemies were gone. I wondered whether I would make it through an ordinary day. Two liters of forty-proof liquor was, more or less, the quota. The Army called the period Terminal Leave, and the name felt accurate: I had no desire to participate in it, and I certainly did not feel eager to survive it.
Eventually, scraping myself out of that endless puddle of vomit, I learned that being broke and hungry could drive me a little harder than guilt, anger, self-loathing and self-defeat. I did not particularly want to survive. Some part of my biology, however, would not let me starve. In my head, I could choose not to eat. In practice, I was incapable of allowing that choice to finish me.
Before the Army, I had been parentless and alone. I knew very little of the world beyond a sequestered life that revolved largely around dairy farming. I did not attend school in the way most children did. At fifteen, I went to college. At seventeen, I joined the Army because it seemed like my only opportunity to have a chance at success. When I came home, I had nowhere to go and no reason to believe the world held anything great for me. I was breathing, but I had no direction, no motivation and no desire for more.
THE COMPUTERS CAME FIRST
Computers were different. I was good at them. In 1995, I built custom computers and paid $495 a week for radio advertising. I had a small business that paid for college. I became one of the first 250 sellers on eBay, even though I had to lie about my age to participate. eBay and computers were what I understood. When the money from leave ran out, I moved into commercial construction. It was good, hard work that required no particular social skill: build a building, and then the building exists. Later came six months of maximum unemployment, self-loathing and selling things online. Selling things online was not a side interest. It was what I lived for. I did not know why it spoke to me, but it was the only work that seemed to take hold of my heart.
Years of building and running job sites eventually made me a licensed engineer-in-training. At the same time, I ran online sales and built websites. That was the shape of my life: build things, sell things, and help other people sell things. Then, almost eight years into that work, I received an email from a browse.com address. I knew what “browse” was, and I was not impressed. I had been focused on the growth of eBay traffic through Google, and I did not need an online amateur to explain the internet to me. The message felt like a trick.
I was also turned off by Jeff Bezos’s public stories about the beginning of Amazon. I remembered the story of a dropshipping operation that did not maintain its own inventory: when an order arrived, the seller bought the required book from a wholesaler, whose minimum order was ten, and used a website glitch to add nine nonexistent books to the purchase. Bezos had described that workaround as part of the company’s origin. The story, along with the early names considered for the company, left me skeptical. I had spent too much time building legitimate online traffic to feel impressed by a beginning I remembered as an exploitation of deceit.
That email nevertheless became one of the most substantial messages of my life. I learned that progress did not always require taking a hard stand. Sometimes the organic flow of a system revealed where energy already wanted to move, and the useful work was removing the barrier rather than forcing a preferred route. The opportunity allowed my systems mind to engage with operations, discard preconceived feelings about an idea and welcome any approach that could improve the metrics without damaging the larger journey.
The people were just as important. I connected with humans across the world, across cultures and languages, and across the different pieces of a product process. Later, the people became a team: people to teach, coach, demonstrate, encourage and help grow. I had always liked math, logic, the penal code and software because they seemed to divide the world into true and false. Amazon taught me that the human being was part of the system, not an error inside it. Human life was messy, and that mess was not a defect to be eliminated. It was where connection came from. Amazon hired me into a remote digital world, and nothing could have brought more connectivity into my life.
THE OFFER
The email was not nonsense. I was being recruited as a seller specialist. Before accepting, I needed to understand more, so I got hold of Fluid Concepts and Creative Realities. I believed, for reasons I could not fully explain, that an answer would present itself somewhere inside that book. Perhaps one did. I took the job, and my mind entered Amazon completely.
I became known for using analogies to teach Amazon processes. I had also spent five years working with artificial intelligence before the company’s AI rollout. I was one of twelve people selected for the global rollout team. It was, as I understood it, the largest global AI rollout and the largest rollout of AI behind a guardrail that had yet occurred. That period carried me through Wainwright, the Wainwright building, Jumbo and Numbo; through breadth-first and depth-first expert systems, cytoplasm, the Eliza effect, SME and ACME; and later through Ractor and the half beard. I became lost in the Goldbach conjecture and the Turing test. The details were strange, technical and intensely alive. They were also a direct continuation of the computer and mathematics world that had helped me keep moving after the Army.
I could not say no to that environment. At fourteen, I had entertained myself with manual calculations of square-round transitions. I had learned C. Now Amazon offered a full immersion in the systems that had always formed the baseline of my thinking. The offer was not simply employment. It was an invitation into a machine whose logic I had already begun using to process the world.
The job itself sounded easy and incredibly fun. When Amazon sent a business owner my way, I would audit the operation and evaluate the intake. The first practical assignment was to help people who already had successful brick-and-mortar businesses make the jump into Amazon. That transition was not as simple as opening Seller Central and uploading products. A local retailer could understand wholesale purchasing perfectly well and still be unprepared for scale, international sourcing, freight forwarding, distributed inventory and the operational requirements of a national marketplace.
I took people and their product ideas all the way upstream: idea, factory, China, port, boat, customs, truck and warehouse. Then I helped distribute the inventory across the country so the product could actually meet its delivery promise. At the time, inventory in a handful of strategic locations could put most of the continental United States within a few days. That sounds almost quaint now, but it meant I learned the system before the product ever touched an Amazon fulfillment center. I learned how an idea becomes a physical object, how the object gets identified, how ownership is represented, how an item enters a catalog, how inventory becomes eligible, how the product crosses oceans, how it clears customs and how it becomes visible to a customer hundreds or thousands of miles away.
THE NEXT CONSTRAINT WAS ALWAYS DOWNSTREAM
Once you improve a process, the part immediately after it becomes more important. That was my career. If inbound becomes better, inventory accuracy matters more. If pick becomes better, pack gets hit harder. If pack becomes better, the dock has to absorb more. If the dock becomes better, the yard and transportation plan become the constraint. If one building becomes stronger, the downstream building can become overwhelmed. If several buildings become stronger, the problem becomes regional flow.
The progression was not simply inbound, pick, pack, dock and flow because those happened to be the next jobs. It was fix this, expose the next constraint, and go there. At some point, the work stopped feeling like career movement at all. It felt like troubleshooting a machine that happened to be hundreds of miles wide.
WHEN A SITE WIN IS A NETWORK LOSS
One of the easiest mistakes in large operations is celebrating a local win that makes the network worse. Imagine a site processing every possible unit and celebrating a great night, only to send the last scheduled trailer away thirty percent full. Someone can look at the building metric and say, “Great job.” If another site in the same region had freight that could have filled the unused capacity, however, the network may have paid thousands of dollars for a victory that existed only inside one building.
The building won. The network lost. I did not want one building to look excellent while somebody downstream was buried. I did not want trailers moving half empty while another site needed capacity, and I did not want one site hoarding equipment because it had found a way to push more volume while another site could not operate. If I optimized one building too aggressively without looking at the rest of the network, I could create exactly the failure I was trying to solve. So I had to move.
THE JOB LOCATION FOLLOWED THE CONSTRAINT
There were buildings where I could increase outbound capability enough that the next problem was physically somewhere else. A downstream sort center could become overwhelmed. Trailer utilization would shift. Go-cart and equipment needs would change. Transportation requirements would change. A building engineered for one level of volume could suddenly operate at multiples of what anyone expected when it was designed.
At that point, staying in the same building was not necessarily useful. The building was no longer the problem. The network was. Site ownership was never the obvious end state for me because owning one building could make the system smaller. Regional flow made the system larger. I liked the larger machine. I liked moving around. I liked seeing how one decision changed everything around it.
There were personal constraints as well. I did not want to relocate away from my children. The career path that might have looked obvious from the outside was never the only thing being optimized. Life is a system too, and the best operational decision is not always the one that produces the most impressive title.
THE 2020 CONVEYANCE
The clearest example of this pattern was a conveyance system I helped engineer in 2020. It was not a normal installation in which a beautiful standardized design arrived with every component selected in advance. COVID had destroyed normal equipment availability. Nobody was manufacturing conveyance on a normal timeline. The practical strategy was to acquire whatever industrial conveyance could be found, sometimes from multiple incompatible systems, and make it behave like one machine.
I was physically and virtually involved in the installation and troubleshooting. Sensors, speeds, diversion logic, stacking areas, container eligibility, dock-door relationships, and the rules governing where boxes were allowed to go all had to agree. The physical machine and the virtual logic had to describe the same operation. It had to run not occasionally but roughly twenty-three hours a day. There is nowhere for a bad theory to hide when steel, software, people and freight must agree for twenty-three hours.
Later, I returned and pushed the system beyond twice its original capacity. Eventually, the operating envelope moved so far that what had once looked like high-volume performance became ordinary. That is what happens when you keep removing constraints. The next one gets louder.
FROM THOUSANDS TO TENS OF THOUSANDS
At one point, a high-volume event around twenty thousand pieces could feel enormous. Later, ordinary days exceeded that. High-volume events moved north of sixty thousand units per day. The numbers matter, but the important question is what the number does to the architecture.
More volume changes staging, labor, trailer requirements, door utilization, downstream sort capacity and maintenance. It changes the amount of freight waiting at every point in the system. It changes which bottleneck matters. You cannot celebrate increased throughput and stop thinking. Every improvement changes the question.
THE BUILDING HAS A SHAPE WHEN IT IS HEALTHY
After enough time inside a system, you stop consciously calculating every variable. You can look at a lane and know it does not look right. You can hear a system and know something is wrong. You can see accumulation where accumulation should not exist. You know when a merge is being pushed too hard, when a downstream area is starving, and when a staging area is beginning to become the next problem.
Some of that knowledge becomes data. Some becomes pattern recognition. Pattern recognition only has value because it was built from thousands of encounters with reality. A feeling is a hypothesis. Then you check it. If the data says your logic is wrong, change the logic. If reality says the model is wrong, investigate. Confidence cannot override known system behavior simply because someone feels like trying something different tonight.
Small deviations propagate. A tiny local change can become hundreds of jackpot packages, a stopped conveyor, a trailer problem, an equipment problem, a downstream capacity problem and somebody hundreds of miles away wondering where the equipment went. That is why architecture matters.
THE BOTTLENECK KEPT MOVING
What looks like a scattered career from the outside makes complete sense to me. I kept fixing what was in front of me. Then the next thing broke louder, so I went there. Sometimes the bottleneck was a process. Sometimes it was a piece of equipment, a conveyance design, staffing behavior, another department, another building, multiple buildings or regional flow.
Later, when physical operations became software, dashboards, standard operating procedures, knowledge systems and applications, it did not feel like a new world. It was still the same set of questions. Where does the thing enter? What owns it? Where is state held? What is allowed through? What happens when capacity is exceeded? Where does it divert? What handles the exception? What does this feed next? What becomes the bottleneck if I fix this?
The rollers disappeared. The logic did not. I never really followed a career ladder. I followed the constraint.