Enzyme is a new LLVM-based differentiation framework capable of creating fast derivatives in a variety of languages. In this talk we will showcase improvements in Enzyme.jl, the Julia-language bindings for Enzyme that enable us to differentiate through parallelism (Julia tasks, MPI.jl, etc), mutable memory, JIT-constructs, all while maintaining performance. Moreover we will also showcase Enzyme's new forward mode capabilities in addition to its existing reverse-mode features.