So I get how this could sound compelling with that framing, but note that:

• This was part of a hack project by 4 students in 2020
• The tool they built in the hack was an ML computer vision system to validate ballots, in an effort to reduce mail-in ballot rejection rates
• As a part of testing this project, they needed a way to generate a large amount of ballots which the system could then validate
• Five years later, one of the authors of this hack works for DOGE

Like, take a look at the code - it's trivial, in large part because it was made by college students in their early academic career. Creating something of similar caliber would be extremely trivial.

That this student hack project would have been used as a part of a greater scheme of election fraud seems highly unlikely.