This article is incredibly misleading in numerous ways. You can find details on pretrial release, like all other issues related to the justice department from the Bureau of Justice Statistics. This [1] paper is extremely relevant.
The normal process in the US is that an individual's case will be seen by a judge within 24 hours of their detention and the judge will assign bail. Contrary to the media, the defendant is often not even present for this process. They will simply be pulled from their cell or detention area and informed what the judge decided. And in the vast majority of cases this is that the individual is released for 'free', or released on personal recognizance. Even in felony cases some 32% of individuals are released without financial condition. When bail is set, individuals can generally get out for 10% of that amount. In some cases they can do it directly through the court, and that 10% will be refunded once the individual returns to court and completes their trial. If not they can go to bail bondsmen who take the 10% as their profit margin, and pay 100% to the court themselves. The bailbonds companies then get their 100% back from the court once the defendant shows up and completes their trial. This is where modern day 'bounty hunting' can come into play -- the bail bondsmen have a strong incentive to ensure that their clients do not flee.
Bail tends to be set in relation to the criminal record of an individual, the seriousness of an offense, how much of a flight risk they are, and whether or not they're likely to be a danger to themselves or others if released. So for instance of all individuals with no prior convictions 77% are released, 63% of those with misdemeanor priors are released, but only 46% of those with felony priors are released. This is also why there are demographic differences in bails and release rates (age, gender, race, etc). Two individuals may commit the same crime, but when one has a criminal record and the other does not - there will be different outcomes. Articles typically try to spin this as prejudice (as this one did), but it is in no way shape or form prejudicial.
There are also statistics that can be used to, inadvertently, test how reliable detention vs release decisions are. In particular when there are emergency releases (such as because of jail overcrowding) where defendants who would not normally be released, are, about 52% end up being charged with some sort of pretrial misconduct. The rate for those who were intentionally released ranges from 27% to 36% depending on the release type. Anyhow, there's an immense amount of other actual information and data on that paper alone, including things like regression comparison of release rates vs reoffense rates to see if the systems in place are justified or not. And in some cases they are not, in some cases they are. I think it's reasonable to conclude that it's a system that does pretty well, but could be made better.
[1] - https://www.bjs.gov/content/pub/pdf/prfdsc.pdf