In a Maple Valley, Washington home, a garage long abandoned for parking has been turned into a workshop filled with old computers, CRT monitors, servers and assorted electronic parts. The shelves hold artifacts spanning roughly five decades of computing — the collection includes more than a hundred complete computers from 1970s minicomputers through 1990s Silicon Graphics workstations to 2000s HP machines. Including video games, software, manuals and other relics, the total inventory exceeds one thousand items.
The collection is assembled and maintained by Darko Mesaroš, an Amazon Web Services (AWS) developer advocate. By day he works as an AWS Distinguished Developer Advocate; earlier in his career he gained sysadmin and engineering experience at AT&T and at Tipico. Photos from the garage were credited to Mariana Mesaros, courtesy of AWS.
Restoring old hardware and giving it new tasks
For Mesaroš, rescuing obsolete equipment is only half the project. He also experiments with using those restored machines for tasks their designers could never have imagined. A Commodore 64 (introduced in 1982), for example, has been used to interact with an AI-based programming assistant. From a 1986 Tandy 102 portable computer he deployed a Kubernetes system. He connected a 1984 DECwriter IV teleprinter to Amazon’s Kiro developer AI tooling so that the AI’s responses were printed on paper while modern cloud software handled processing in the background.
One prized item in the collection is a bright-blue Silicon Graphics Indy workstation; Mesaroš says it would be the first thing he rescued if the garage caught fire. Restoring decades-old equipment often means there is no ready-made solution or easily available documentation, which forces patience, focus and creative troubleshooting.
Bridging eras with standards and tools
Mesaroš’s work highlights that connections can be built between different eras of computing, and his employer AWS points to the role of standard communication protocols and modern developer tools in making that possible. The process of repairing and adapting old hardware reinforces skills — diagnosing failures, reverse-engineering, and improvising — that Mesaroš believes will remain valuable even as AI automates more development tasks.
His personal interest in vintage computers goes back to childhood in Serbia. Although computing advanced rapidly during the 1990s, Mesaroš recalls that between 1991 and 2003 the Commodore 64 was his primary exposure to the digital world. As an adult he sought out machines from eras he had missed, starting with a few Sun workstations and expanding the collection from there.
What developers should keep in mind
Mesaroš warns that while AI can now generate code and create applications from simple prompts, relying uncritically on AI suggestions risks eroding deep understanding. He argues the most successful future practitioners may not be the fastest coders but those who remain curious, dive deeply into problems and can independently evaluate the solutions AI proposes.
The Maple Valley garage functions as both a technological time capsule and an experimental lab: vintage machines come back to life, and modern AI and cloud technologies are tested in unusual contexts. Mesaroš’s project is a reminder that progress in computing is driven not only by more powerful hardware and new algorithms, but also by the curiosity that motivates someone to learn how a machine actually works.
