Only auditing the paper trail when the electronic results seem suspicious is a vulnerability in my opinion.
All these technical solutions target the non-problem of counting results quickly. You can scale up paper ballot counting by including more people, which also strengthens trust in your democracy.
Basically because of ballot stuffing, booth capturing, ballot tampering, fake ballots etc etc. Paper ballots have been a major vurnerebility from experience of several decades in rough and tumble politics. The article on EVMs I linked has descriptions of it. The EVMs rate limit voting to prevent this from being a major motivation for political parties. The approach that seems to be tried and tested and working well seems to be to use dumb unconnected and robust electronic voting machines which can't be hacked without physical possession and the using massive security to prevent physical possession to attackers.
The problem is they have no protocol for better protecting the ballot box and responding to ballot stuffing attempts and incidents. If you're not going to protect the ballot box it doesn't matter what tech you use for voting.
I maintain that paper ballots are the best voting mechanism. They are cheap, prevent intimidation, scale and reasonably trustworthy.
I say they scale because you can parallelize easily by getting more citizens counting ballots. This is a good thing. The more people involved in counting the more you have to coerce to reasonably affect the outcome.
Electionic voting machines like in the US seem insane to me. All the power is in a select few who create the machines.
If you can't source enough trustworthy volunteers to count paper ballots then your democracy might be screwed anyways.