Writing a just-in-time (or JIT) compiler from scratch is hard work. Using the LLVM framework to build a JIT, however, is an obvious choice - it offers impressive performance without the need to develop a compiler from the ground up. Despite these advantages, many virtual machines still rely on custom-built compilers. The reason? LLVM's performance gains often come at the cost of lengthy compile times, which can slow down application warmup and increase CPU load. At Azul, we have utilized LLVM for our JIT, especially since our focus on large, long-running servers has made warmup time less of a concern. However, the increasing prevalence of small containers has driven us to rethink our warmup strategy. In this talk, we will share how we have successfully reduced warmup times while maintaining the benefits of LLVM as our compiler. Spoiler alert: the solution involves even more LLVM.