Pyodide: Running Python Data Science & NumPy Directly in the Browser via WebAssembly

Pyodide ports the CPython interpreter and the scientific Python stack (NumPy, Pandas, SciPy, Matplotlib, scikit-learn) to WebAssembly (WASM) via Emscripten. This allows web applications to execute complex Python data transformations, statistical models, and interactive sandboxes 100% client-side inside the user's browser with zero backend compute cost.

1. Loading Pyodide Inside a Dedicated Web Worker

Because initializing the WASM runtime and executing heavy array computations can block the main UI thread, production applications should always host Pyodide inside a background Web Worker:

// pyodide.worker.js
importScripts("https://cdn.jsdelivr.net/pyodide/v0.26.2/full/pyodide.js");

let pyodideReadyPromise = (async () => {
  self.pyodide = await loadPyodide();
  await self.pyodide.loadPackage(["numpy"]);
})();

self.onmessage = async (event) => {
  await pyodideReadyPromise;
  const { id, pythonCode } = event.data;
  try {
    const result = await self.pyodide.runPythonAsync(pythonCode);
    self.postMessage({ id, result: result?.toJs ? result.toJs() : result });
  } catch (err) {
    self.postMessage({ id, error: err.message });
  }
};

2. Zero-Copy Array Sharing Between JavaScript and Python

Using pyodide.toPy() and typed array buffers (Float32Array), you can pass megabytes of telemetry or audio samples between JavaScript and NumPy without JSON serialization overhead.