The seams are the job now

I've spent twenty years watching hiring managers flatten a career into a single noun. Backend. Frontend. Manager. The résumés that scanned were the ones that could be read in a second. Mine never scanned. I used to treat that as a problem I had failed to fix. I don't think it is a problem anymore.

Specialization paid when execution inside a discipline was expensive. A specialist could produce more, faster, with fewer mistakes, and that difference was worth a headcount line. Execution got cheap in several disciplines at once, and the expensive part moved to the connections between them. That is where I have actually lived.

The résumé that reads as unfocused

Here is the list, without the usual cleanup. Seven years at UPS in Phoenix: district software engineer, tech support, supervisor, deployments across three states and hundreds of locations. Sole corporate PHP developer at Fidelity National Financial, the largest title insurance company. Project manager at Electric Sheep for builds shipped to IBM, AOL, STA Travel, and NBC, which was also my first PHP and MySQL API work. Content lead for Seesmic's corporate blog. Then the campaigns: McCain in 2009 and 2010, Newt in 2012, Romney in 2012. Senior technical lead at Zibtek from 2014 to 2017, delivering for clients including Adobe and Livefyre with an engineering team in Pune. Individual contributor to CTO at Life Happens, 2018 to 2023. Under all of it, a consulting practice, Vaught Consulting and then Agency Intentional, running alongside the jobs rather than instead of them.

There is also a rug tufting studio, Common Thread, that I co-founded with two partners in downtown Xenia, Ohio. It is a live testbed for applying AI to a local service business, and it belongs on this list for one reason: a shop floor, a booking calendar, and customers who walk in are an operation, and applying AI to an operation is seam work. You cannot do it from a stack preference.

In 2016 a hiring manager reading that list for an engineering role was looking for the plot. Supervisor, then sole developer, then a blog, then campaigns, then Pune. The charitable reading was range. The actual reading was that I had not picked a lane. I understood that reading. I would have made the same call.

What the list means now is simpler. Each job put me inside a messy operational problem, and I had to understand how the people inside it worked, then build something they could use. That is the only through-line that is true.

The specialization trade

Depth paid because execution was expensive. If you needed a Laravel application, a specialist who had only done Laravel beat a generalist who had also run a blog and a campaign. Companies organized around that fact: ladders rewarded narrower excellence, job posts asked for years in one stack, interview loops tested depth because depth was a decent proxy for production. That was not foolish. It was a cost structure. When typing out a working system took weeks, you hired for typing.

What it cost is less often named. The connections between disciplines went unowned. Somebody had to stand in the doorway, and that somebody looked, on a résumé, like a person who could not commit. I held two titles at once at Life Happens, Director of Engineering and Director of Customer Success, because the loop between what shipped and what members experienced was the actual job.

What changed

Execution got cheap in every discipline I know, at roughly the same time. You can get a first draft of the code, the copy, the spec, and the support reply without hiring a specialist for each. The draft is not the job. The job is knowing whether the draft is wrong, and wrong in a way that shows up at a seam: between engineering and support, between a campaign site and a rule that cannot move, between a shop's calendar and the model that thinks it understands the calendar.

Value moved to the seams, and the seam is where the generalist has always lived.

This is not a claim that specialists are obsolete. A generalist who cannot go deep in anything is a tourist, and the failure mode is the person who is shallow in all of them. Range is not depth in several places, and it is easy to confuse the two if you are the one telling the story. The useful shape is someone who has been deep enough in more than one room to recognize the handoff, and who will own the handoff instead of throwing it.

A 2016 hiring loop was designed to reject that person early. I think the signal flipped.

How I would actually interview for this

I would stop asking for a single noun, and stop using a whiteboard puzzle in one language as a proxy for whether someone can tell when a generated answer is lying. I would ask them to walk me through a system they did not invent, and what they changed after they understood the people who used it. I would ask where two functions handed work to each other and what broke at the handoff. I would ask for a discipline they are not credentialed in that they still had to be competent in last year, and listen for whether they can describe the actual work or only the vocabulary.

The tell for the shallow generalist is speed without friction. People who have really crossed a seam remember the part that did not fit.

I ran hiring for an engineering organization, and I do not have a clean rubric for this. I am suspicious of anyone who does. Distinguishing range from restlessness is the whole problem. The only thing I am sure of is that the old scan, the one that wanted a single noun, now filters out exactly the people the work needs.

Related: Many hats, Player-coach