How doe Maximum Likelihood estimation work in Logistic regression




While learning about MLE I came across:

Does this mean that for logistic regression MLE works such that for each record the probability of observing the outcome is maximized or does it mean that the MLE tries to find the values of betas such that the probability of observing the true proportion of 1’s in the sample is maximized.
That is if in the sample the proportion of 1 is 70% MLE tries to find those values of betas for which the the predicted proportion is as close to the true prop ??
Or does it find betas for each record and then average to give the betas of the logistic regression model??
Can someone please clarify on this??