About
Steve Dolinsky
Technical program manager working in hardware NPI, with a background across cloud infrastructure, data centers, and server platforms. I write here about what the discipline becomes when coordination itself is programmable.
I've worked in tech, data centers, and cloud computing since January 2018, and in technical program management since October 2019. For several years I split my time between cloud infrastructure and the physical server platforms underneath it. That kept me on the seam between hardware and software, where a plan has to survive both a code review and a loading dock.
Now I'm fully focused on hardware new product introduction. It's an amazing vantage point for this inflection in tech. AI can move the software layer quickly; new hardware still has to make it through design, qualification, manufacturing, supply chains, and a loading dock. Earlier I led cloud platform transformation work through a major platform acquisition and ran mission-critical data center migrations. The through-line is programs where a mistake cannot be fixed by editing a slide. I hold a B.S. in Information Technology Management.
This site exists because of a question I keep running into from both directions: how far a program's reported state can drift from what's actually happening, and how much of that gap is a design problem rather than a communication problem.
What I'm working out here
The starting point was simulation bias. Reading Jean Baudrillard's Simulacra and Simulation on vacation gave me language for a program failure I already knew: the status model can detach from the work, then become the reality the organization acts on. AI raises the stakes because the model can now produce the account and take action against it.
Everything else here follows from asking what to do about that. This site explores whether technical program management is evolving toward systems engineering for organizational execution: designing how commitments are represented, observed, governed, and acted on without confusing the representation for reality. The thesis page is the full version.
Program Operating System is my provisional name for the architecture that has emerged from that question. It is a working model, not a validated category. The method is simple: build, break, measure, revise, reproduce, publish.
Contact
The most reliable way to reach me is LinkedIn. I'd genuinely rather hear that a piece of this is wrong than that it resonated.
New writing goes out over RSS. There's no newsletter yet and no tracking beyond aggregate page counts.
The usual disclaimer
Everything here is published in my personal capacity. The opinions, models, and arguments are mine and do not reflect the views of any past, present, or future employer. I discuss work in general terms so that nothing confidential, proprietary, or privileged is disclosed, and nothing on this site should be read as an employer endorsing any of it.