I've worked with a large Cassandra cluster at a pervious job, was involved with a Scylla deployment, and have a lot of experience with the architecture and operations of both databases.

I wouldn't consider Cassandra/Scylla a replacement for Postgres/MySQL unless you have a very specific problem, namely you need a highly available architecture that must sustain a high write throughput. Plenty of companies have this problem and Scylla is a great product, but choosing it when you are just starting out, or if you are unsure you need it will hurt. You lose a lot of flexibility, and if you model your data incorrectly or you need it in a way you didn't foresee most of the advantages of Cassandra become disadvantages.