Numba's JIT Compiler Now Runs Scientific Python Right in Your Browser
Numba, the popular just-in-time compiler that speeds up numerical Python code, has been ported to run inside JupyterLite—the browser-based version of Jupyter notebooks. Normally Numba relies on LLVM to compile Python functions down to fast machine code on your local machine or server. This project gets that same compilation pipeline working inside WebAssembly, meaning the JIT compiler itself runs in the browser sandbox.
That's a nontrivial engineering feat: LLVM and Numba weren't designed with WebAssembly constraints in mind, so getting them to cooperate involves work on packaging, memory limits, and toolchain quirks. The payoff is a scientific Python stack (think NumPy-style workloads) that can execute with near-native speed without a server, a container, or a local Python install—just a browser tab.
This matters for teaching, demos, and reproducible research where asking someone to install a Python environment is a barrier.