Instead of:

def append_and_recurse([h | t], m), do: append_and_recurse(append(t, f(h)), m-1)

“recurse with t and then f(h)”

def recurse_then_reverse([h | t], m), do: recurse_then_reverse([h | t], [], m)
def recurse_then_reverse([h | t], rest, m), do: recurse_then_reverse(t, [f(h) | rest], m-1)
def recurse_then_reverse([], rest, m), do: recurse_then_reverse(Enum.reverse(rest), m)

This keeps the list of “things to evaluate next” in reverse order until the first list is empty, then uses a single Enum.reverse.

Take a look at the Erlang stdlib’s :queue module for references and more examples of this approach.