Can you write a performant string processing scripts in Elixir? This talk attempts to answer that question while incrementally optimizing a simple implementation of a script, introducing many interesting parts of Elixir on the way, including ETS, iolists, compiled binary patterns, and concurrency. With a little bit of knowledge, we end up with a 10x speedup over the original implementation, and Elixir becomes a viable and performant scripting language, and this also carries over into most data processing tasks.

But what are the risks and trade-offs when optimizing, and how do you do it in a productive way?