This has me really curious about the state of software engineering principles in academia if you can find bugs that invalidate 18 months of work.

I'm not saying it should have been preventable. It just looks like there may be opportunity to improve practices.

Admittedly I'm completely naive to the domain. Are there no forms of validation checkpoints you can reach where your foundations are rock solid and well backed with tests and such?

I had a Prof in grad school who lost years of cryospheric research data because an external hard drive was stolen. This was in 2010. It was a head scratcher, especially being faculty-adjacent to some of the best CS and engineering faculties in Canada.