I see you have a bunch of good answers now, so I’ll ask; if you are comfortable self hosting, why not consider a VPS? Yes it can be a little bit of maintenance, but it’s very minimal and you get far more flexibility and the ability to further develop those selfhosting chops.
I use sendgrid as my outgoing smtp relay to avoid ip reputation issues you mention. You still have to configure your dns settings for spf and dkim pointing at their servers instead of yours. Their free tier is 10x the email I’ll ever send so it doesn’t cost anything. There are a few companies in this space with free tiers. It works, but it isnt Gmail level deliverability. I still get spam binned occasionally.
This is signal detection theory combined with an arms race that keeps the problem hard. You cannot block scrapers without blocking people, and you cannot inconvenience bots without also inconveniencing readers. You might figure something clever out temporarily, but eventually this truism will resurface. Excuse me while I solve a few more captchas.
Data scientist here; there simply are not enough murders to model this, so they will need to use proxies for “likely” murderers (like any sort of violent crime). That means the model will very strongly target people who are over-policed (minorities) and those more likely to actually get caught and charged for things, and thus be in the training data set (poor people). It will also fail spectacularly for this purpose because even a highly accurate model will produce almost 100% false positives -again, because actual murders are so vanishingly rare. The math just doesn’t work.