Welcoming Tom Kovalcik, PE, Our Founding Structural Engineer

Tom Kovalcik, PE has joined Structured AI as our Founding Structural Engineer. He has spent over ten years in construction and structural engineering, and he has worked all three sides of a drawing set: the contractor building it, the engineer sealing it, and now the ML engineer teaching a model to check it.

That combination is rare, and it is exactly what an AI design reviewer needs. A model that checks structural drawings has to know the difference between an error that is cheap to catch and one that surfaces as a change order six months later. Tom has been on the receiving end of both.

From the field

Tom started as a project manager for Nan, Inc. on three light rail stations of the Honolulu Rail Transit Project. He administered roughly $10M in subcontracts, ran fabrication and erection of over 1,000 tons of structural steel, and wrote more than 300 RFIs.

Writing 300 RFIs is an education in itself. Each one is a question the drawing set failed to answer. After a few hundred, you stop reading a set as a stack of sheets and start reading it as the relationships between them, which is exactly the view an automated review has to take.

To the design side

Then Tom crossed over to design. Six years between Kimley-Horn and IDA Structural Engineers, and about 50 projects spanning transit, aviation, municipal, EV, solar, water, roadway, and building work. Steel, concrete, wood, and masonry. CBC and IBC.

Some of it carried real consequence. He ran analysis and failure investigation on the PHX SkyTrain expansion joints, on steel already carrying trains. As West Coast regional manager for Kimley-Horn's temporary structures program, he handled peer reviews and field inspections for more than 40 concerts and festivals a year, on stages that go up and come down inside a week.

Day to day, that meant running five to ten jobs at once as a PM while mentoring junior engineers and reviewing their work. He holds a PE in California and a master's in structural engineering from UC Berkeley.

He has reviewed enough drawings to know which errors are cheap to catch, and which ones show up as a change order six months later.

And into the model

Toward the end of his time in design, Tom started writing his own Python scripts and Mathcad templates to get through the repetitive calcs faster. That instinct is what eventually brought him here.

He left to take a master's in data science and AI at the University of San Francisco, built computer vision pipelines for spotting roadway assets in field imagery, and then came looking for us.

What Tom will own here

Tom will lead our structural engineering practice: building out the structural code compliance and cross-disciplinary QA/QC checks in our platform, and working directly with engineering firms to make sure our reviews reflect how structural teams actually work.

We have always built this alongside practicing engineers rather than around them. Having a licensed PE who also writes computer vision pipelines inside that loop shortens the distance between what a senior engineer would flag and what the platform flags.

Glad to have him on this side of the table.

Better drawings build a better world.