"Hand-coded" may be an exaggeration, depending on your field.
ML systems need good training data (where "good" might mean "extremely voluminous") and careful tweaking of parameters. I think an ML expert would call a system where a substantial amount of time is spent here "hand-coded".